Student

Student--Faculty Programs
2015 Abstract Book

STUDENT-FACULTY PROGRAMS
2015 Abstract Book

This document contains the abstracts of the research projects

conducted by students in all programs coordinated by Caltechs
Student-Faculty Programs Office for the summer of 2015.

Table of Contents
Summer Undergraduate Research Fellowships (SURF)

WAVE Undergraduate Research Fellowships

91

Amgen Scholars Program

97

Laser Interferometer Gravitational-Wave Observatory (LIGO)

103

SURF
SUMMER UNDERGRADUATE
RESEARCH FELLOWSHIPS

S
U
R
F

Reference Frames Employed by Neuronal Populations in the Posterior Parietal Cortex of Humans
Minaam Abbas
Mentors: Richard A. Andersen and Tyson Aflalo
A brain machine interface (BMI) works by decoding the intentions of the user from patterns of neural activity. In
this study, we seek to understand the reference frame of intention signals in the posterior parietal cortex of a
tetraplegic human participating in a BMI clinical study e.g. whether intention signals are represented with respect
to where the subject is looking, the current location of the effector, or the body or world.
Neural population activity was recorded from an electrode array implanted in the PPC of a tetraplegic subject
during a modified delayed reach task. We instructed different gaze and starting effector positions to understand
how these factors influenced neural firing for intended reaches. Analysis was performed on single units as well as
on the whole neural population recorded. We found that both the initial hand and eye positions affect intention
decoding during reaches providing evidence that understanding neural reference frames is critical for optimizing
BMI performance. Preliminary results suggest that individual neurons can code reach intentions in a variety of
reference frames providing a rich set of intention related activity. The results help identify the regions of the brain
important for spatial transformations underlying movements. Additionally, the findings provide critical information
about how future decoding algorithms should interpret signals when used in a BMI.
Inner Core Translation and True Polar Wander
Jacob Abrahams
Mentors: David Stevenson and Hao Cao
Recent seismological observations point to asymmetric seismic velocity distribution in the Earth's inner core, which
has the potential to cause long term East-West asymmetry observed in the Earth's magnetic field. Data from the
MESSENGER spacecraft show that Mercury's magnetic field is strongly asymmetric in the North-South direction yet
lacks East-West asymmetry. Current understanding of the Earth's inner core's asymmetry, in particular the inner
core translation scenario, tend to favor the asymmetry in the direction normal to the spin axis when the heat
transfer efficiency is isotropic near the inner core boundary. \newline
Here we explore the density distributions and moments of inertia (MoI) of a translating planetary inner core. In
particular, we focus on the relative magnitude of the inner core's polar moment of inertia and equatorial moment of
inertia. We find that in an East-West translating inner core, there is about 10\% probability that the equatorial
moment of inertia can becomes larger than the polar moment of inertia. True polar wander of an internally driven
translating inner core would then rotate an originally East-West translation into a North-South translation. We
propose that a North-South translating inner core could produce a magnetic field which qualitatively resembles that
of Mercury's, given that the inner core is viscous enough and can true polar wander. In addition, the high likelihood
of the translation direction establishing a maximum moment of inertia normal to itself likely explains Earth's
equatorially aligned translation.
Graphical User Interface (GUI) for MacroDFT and Parallel Computing
Oreoluwa Adesina
Mentors: Kaushik Bhattacharya, Mauricio Ponga, and Dingyi Sun
Bhattacharyas group developed a software called MacroDFT. This is a large-scale Density Functional Theory (DFT)
software that requires input from its user. But unfortunately, this software did not have a graphical user interface
to accept the users data, and this proved a drawback in the user experience. The aim of my project was to create
a graphical user interface which is able to collect user input for the code. Using the Python programming language,
a graphical user interface (GUI) was designed and after a few weeks, the GUI became a complete standalone
windows application which solved the problem that slowed down the progress of MacroDFT.
In addition to this project, I worked on another project with the aim of reducing the amount of time computers
spend in performing work assigned to them. Using MATLAB, I performed simulations which tested different
algorithms. This project I still a work in progress but the intention is to obtain the most efficient algorithm which
simultaneously saves the most time, and then apply it to computers and get work done quicker and better.
Global Analysis of Rotational Spectra of the Ground State and Excited Torsional State of
Propylene Oxide
Avinash Agrawal
Mentors: Geoff Blake, Ian Finneran, and Jacob Good
This paper analyses the spectra of Propylene Oxide with special attention to the splitting caused by the methyl
group internal rotation. This molecule is important as it a candidate for first chiral molecule found outside our solar
system. Previous data are used alongside more accurate data collected in the Blake lab from 8 - 18 GHz to refine
the values of the inertial parameters, the internal rotation barrier height, and the rotational distortion constants.
Multiple programs are used for ab initio calculations and experimental data fitting such as SPFIT, XIAM, and

Gaussian 09. Ab initio calculations indicate that the barrier height is 915 cm-1 and the rotational constants are
A: 18094, B: 6719, C: 5990, all in MHz. Subsequent work will include the use of lab data as well as previous data
to modify the ab initio calculations and to fit ground state and excited torsional states.
Frustrated Lewis Pairs as a Route for N2 Reduction by H2
Saaket Agrawal
Mentors: Jonas C. Peters, Jonathan Rittle, Miles W. Johnson, and Sidney E. Creutz
The reduction of nitrogen (N2) to ammonia (NH3) is an important transformation relevant to life. Our group has
previously synthesized tripodal iron complexes that can catalytically reduce N2 to NH3. Unfortunately, these
transformations require the use of a very high energy reducing mixture of KC8 and HBArF4, limiting the systems
practicality. A sustainable route for the reduction of N2 to NH3 with a homogenous catalyst will use a much cheaper
and easy to access reductant, such as hydrogen gas (H2). Frustrated Lewis pairs (FLPs), sets of Lewis acids and
bases that remain unquenched due to steric encumbrance, have been shown to be a promising route for activation
of H2. Herein, we present progress towards the utilization of the FLP concept to activate H2 for the reduction of N2
to NH3 on our previously reported tripodal iron complexes. Studies are ongoing.
Observation and Reactivity Studies of a RhI Intermediate in H2 Evolution Catalysis
Luis M. Aguirre Quintana
Mentors: Harry B. Gray, James D. Blakemore, Jay R. Winkler, and Bruce S. Brunschwig
Understanding mechanisms of hydrogen evolution is important for improving catalysts that generate clean fuels.
Catalysts often operate via unknown mechanism(s) due to the difficulty of observing or isolating key intermediates
in the 2e/2H+ reduction process. For our study, we prepared a family of RhI complexes bearing the
pentamethylcyclopentadienyl (5-Cp*) ligand in addition to derivatives of 4,4-bipyridyl or 1,10-phenanthroline
(bound in the 2 mode). Under carefully chosen conditions, these compounds formed intermediate species that
suggest a previously unknown mechanism of H2 evolution. This intermediate complex could be isolated in a MeCN
solution, and even as a solid, making it amenable to further mechanistic investigations. The intermediate species
were analyzed by mass spectrometry, 1H NMR, and 2H NMR (NMR is nuclear magnetic resonance spectroscopy).
Results will be discussed concerning the properties and reactivity of these active intermediates.
Rapid Prototyping of Moderate Complexity Biomolecular Circuits: Bistable Switch
Enrique Amaya
Mentors: Richard Murray, Clare Hayes, and Vipul Singhal
Synthetic biologists adapt engineering principles to biological systems by developing robust and standardized
modules in order to design complex biomolecular circuits. These are used to program cells with desired functions
that attempt to address societal challenges such as deadly diseases and damaged environments. However
implementing a complex biocircuit in vivo is a laborious process as there are long steps between DNA assembly,
transformation, growing and selecting cultures. This can be addressed by using modeling and prototyping to predict
what conditions are required for proper performance. In this project our prototyping device was an E.coli based
cell-free expression system called TX-TL (translation-transcription). With this tool we prototyped several sets of
repression systems, including the promoters pLac and pTet, the repressor proteins LacI and TetR and the inducers
IPTG and aTc. The data obtained was used for parameter estimation using a TX-TL simulation environment in
MATLAB and the lsqcurvefit method. As a final step we plan to build a bistable switch, first simulating the circuit in
silico and then comparing its behavior in the TX-TL expression system to find out if our modeling toolbox reliably
predicts biocircuits performance.
Self-Assembly of Large Structures in Space Using Lagrangian Mechanics
Carmen Amo Alonso
Mentor: Michael Ortiz
In-orbit assembly of structures is a costly and time-consuming task at present. The approach carried out in this
research relies on the invariance of the Lagrange mechanics equations under change of time sign. With this
strategy, the most natural self-assembly mechanisms are easily found, which allows an innovative and efficient
way of self-assembly that has not been studied before. The work focuses on the dissolution of a rotating structure
in a spiral shape composed by discrete units held together. The structure increases its angular velocity until the
peripheral units start to detach as an effect of the centrifugal acceleration. As the structure losses mass, its angular
velocity increases, so it can result in the complete dissolution of the spiral. For this to happen the rate of emission
needs to be stable, what can be reached by imposing the laws of conservation and some design requirements. Just
by changing the sign of time, the way to provide the discrete components and the appropriate geometry of the
structure in order to assure the continuous assembly is reached.

Practical Implementation of Interference Alignment for Cellular Networks

Alexander Anemogiannis
Mentor: Babak Hassibi
Many researchers have proposed various approaches to Interference Alignment (IA), a way of mitigating the effects
of interference in multi-user networks, but these approaches tend to take place in ideal settings somewhat
removed from practice. We present a practical approach that uses minimal channel state information and
encourages fairness among users. Rather than relying on full channel knowledge, we focus on the topology of the
network and represent channels between users with one bit to distinguish between strongly interfering users and
weakly interfering users that can be ignored. Given the topology, we use an alternating minimization between
precoding filters at the transmitters and receiver shaping filters at the receivers to isolate desired signals from
strong interferers. Maximizing the sumrate of the users leads to better individual rates for users with favorable
channel states at the expense of users with poor channel states. In order to encourage fair individual rates, we
instead maximize the minimum user rate, which is a quasi-convex problem that can be solved using the bisection
method.
Development and Characterization of a Josephson Parametric Amplifier for Study of Near Ground State
Quantum Systems
Alexander V. Anferov
Mentors: Oskar Painter and Johannes Fink
Measurements of solid-state quantum systems at or near the quantum ground state need to be preceded by
amplification, since the energies of signals of interest in the microwave range are too small for conventional
detection. Interest in high fidelity measurements of individual photons, which are easily overpowered by
amplification noise, call for robust, high accuracy amplification systems with detection thresholds approaching the
quantum limit. Recent advances in very low noise signal amplification devices have focused on parametric
amplifiers, which offer a very high signal to noise ratio over a very narrow band of frequencies. In order to
maximize the tunability and dynamic range of operation, we develop a Josephson Parametric amplifier, which uses
an array of superconducting quantum interference devices (SQUIDs) composed of superconducting Josephson
junctions to provide additional tunability and the necessary nonlinearity for parametric amplification. This work
details the modeling, development and full characterization of a microwave range Josephson Junction Parametric
amplifier in terms of its frequency tunability, Josephson junction characteristics, drive power regimes, signal
amplification regimes, and gain performance.
Luminescent Ruthenium Probes for DNA Mismatches
Sarah Antilla
Mentors: Jacqueline Barton and Adam Boynton
The development of cancer diagnostics and treatments is a growing area of research. The accumulation of DNA
mismatches is associated with several forms of cancer, and we are interested in designing ruthenium complexes to
function as diagnostic probes for mismatches. The complex [Ru(bpy)2dppz]2+ (bpy= 2,2-bipyridine; dppz =
dipyrido[3,2-a:2,3-c]phenazine) is a DNA light switch because it does not luminesce in aqueous solution but
does luminesce when bound to DNA. Although this complex does luminescence more brightly in the presence of
mismatched DNA relative to well-matched DNA, it is not mismatch-specific. We are aiming to achieve mismatch
specificity by synthesizing [Ru(L)2dppz]2+ derivatives bearing more sterically expansive ancillary ligands (L) to
discriminate against binding to well-matched sites. We characterize these complexes via mass spectrometry, NMR,
and UV-Visible spectroscopy, and use luminescence spectroscopy to measure their emission intensity in the
presence of well-matched and mismatched DNA. We have synthesized two complexes, [Ru(Me4phen)2dppz]2+ and
[Ru(Me2bpy)2dppz]2+, which bear methyl substituents on their ancillary ligands. Compared to their respective
parent [Ru(phen)2dppz]2+ and [Ru(bpy)2dppz]2+ complexes, these new methyl derivatives exhibit significantly
greater luminescence in the presence of mismatched DNA relative to well-matched DNA, suggesting that they may
be used as novel probes for DNA mismatches.
Lepidoptera Identification via Pose-Normalization and Convolutional Neural Networks
Richard J. Antonello
Mentors: Pietro Perona and Steve Branson
A major goal of computer vision research has been to provide efficient and accurate algorithms for the automation
of subcategory recognition. We propose a generalized methodology for the full implementation of fine-grained
classifiers for biological organisms, which we apply to a Lepidoptera test domain. After dataset collection, we use
crowdsourcing tools to fully annotate the collected images. Utilizing redundant annotation of images, we obtain
ground truth annotations for our dataset. With these ground truth annotations, we normalize the images from our
dataset into several standardized poses, which we use as feature vectors to train and finetune on VGGNet, a stateof-the-art deep neural network architecture. In addition to our CNN methodology, we also explore and implement
an alternative technique, based on adaptive kernel density estimation, for the training of a probability map of
species likelihood given geographic sighting data.

Designing Biomolecular Temperature Sensors

Divyansh Apurva
Mentors: Richard Murray, Shaunak Sen, and Dan Siegal-Gaskins
RNA Thermometers are temperature-sensing elements present in mRNA in an organism that have a property of
changing their secondary structures in response to temperature changes. Their location in the 5-UTR regions
genes enable them to regulate gene expression with shifts in temperature. Previously, frameworks for designing
simple synthetic RNATs with different temperature profiles have been reported and a toolbox of RNA thermometers
based on that framework has been constructed. However, the behaviour of the RNA thermometers in live cells and
in cell free systems has not yet been fully analysed and understood. To address this issue, we aim to standardize
and characterize changes in expression for various RNA Thermometers with temperature. With this characterization
and tuning as a basis, a temperature-regulated synthetic biomolecular gene expression circuit will be produced.
The results should present a huge library of temperature-dependent RNA Thermometers which can be used to
increase robustness or sensitivity of bio-molecular circuits.
Virtual Proof of Position of a Living Organism
Fabio Kenji Arai
Mentors: Changhuei Yang and Roarke Horstmeyer
A virtual proof of reality is a new cryptography protocol that goes beyond classical security keys and utilizes more
general physical statements. In this approach, the statement to be proved is whether an organism is dead or alive.
In order to do this, a living organism is confined inside a box made of a scattering material. When illuminate by a
coherent light source, this box will create a unique random pattern (a speckle pattern), which may be detected by
a camera. This speckle pattern depends upon the position of the living organism. By recording speckle patterns
specific to certain positions, it is possible to prove to a verifier that the organism moved, and therefore it is alive.
We have shown the feasibility of this approach using a drosophila melanogaster as our living organism. These
results along with previous work on different approaches shows promise for this new protocol, increasing its
viability for practical applications.
Reciprocity Laws on Arithmetic Surfaces
Diana Alexandra Ardelean
Mentor: Xinwen Zhu
An arithmetic surface is a two dimensional scheme of finite type over the ring of integers. It could be an algebraic
surface over a finite field, or an integral model of an algebraic curve over a number field. We investigate the
possibility of introducing new symbols for arithmetic two-dimensional local fields, such as (()) and use them to
prove various reciprocity laws for arithmetic surfaces. Our study will start from the perspective of describing a
central extension of by fields such as and using it to obtain the tame symbol as the commutator of the
lifting of two elements from to this central extension.
Computational Discovery of Binding Poses for Beta-Ionone to hOR5A1
Naveen Arunachalam
Mentor: William A. Goddard, III

Understanding the mechanism of olfaction in humans is a difficult process due to the computational complexity of
predicting structures for human olfactory receptors (hORs). However, methods previously developed by the
Goddard group have made the problem computationally feasible through GEnSeMBLE, a suite of structure
prediction methods for G-protein coupled receptors (GPCRs). Using GEnSeMBLE, we have found three low-energy
structures for hOR5A1 based on homology to hOR5BC, hOR2J3, and hOR7D4. Using the structures, we have found
three energetically favorable docking poses for beta-Ionone (a typical ligand for hOR5A1). Further analysis of the
results will involve molecular dynamics simulations at 6000 K near the binding site to assess the validity of the
discovered poses.
Lightweight, Adaptable Manipulator for Aerial Door-Opening Rescue Robot
Sarah Keiko Asano
Mentor: Hideyuki Tsukagoshi
My research was to design a 3 degrees-of-freedom, lightweight, doorknob-grasping manipulator actuated and
powered by pressured plastic tubes. This allows more ways of manipulating objects for the operator or program
controlling the finger, but specifically allows a door-opening aerial robot to securely grasp the two main types of
doorknobs, as well as a variety of other objects. The optimum actuation sequences are determined by testing the
security of the manipulator's grip, and an method to grasp arbitrary graspable objects using this manipulator is
proposed.

Protein-Catalyzed Capture Agents Targeting Misfolded Superoxide Dismutase 1

Beatriz Atsavapranee
Mentors: James Heath and David Bunck
Amyotrophic lateral sclerosis (ALS) is a neurodegenerative disease that affects motor neurons in the brain and
spinal cord. Familial ALS can be caused by mutant isoforms of superoxide dismutase 1 (SOD1) that lead the protein
to misfold and adopt toxic conformations. Over 180 mutations in SOD1 are linked to the disease, making it a
challenging therapeutic target. The objective of this project is to develop protein-catalyzed capture (PCC)
agents that target SOD1 mutants. We used an epitope-targeted screen to identify oligopeptides that bind regions
of the protein destabilized upon mutation and an internal fragment revealed upon misfolding. We anticipate that
the PCC agents against the internal fragment and precleared against the electrostatic loop will bind misfolded SOD1
selectively, while the PCC agents against the electrostatic loop will detect folded SOD1. This will allow us to
discriminate between the folded and misfolded species. Screens against one external epitope yielded several
hits whose ability to recognize SOD1 is under investigation. Ultimately, these ligands might be used to mitigate
aggregation of mutant isoforms and treat ALS.
Genes of Trunk Neural Crest Pathways
Caroline Atyeo
Mentors: Marianne Bronner and Christina Murko
Neural crest cells are a multipotent, migratory set of cells that arise in the primitive neural system during
embryogenesis. The cells of the trunk neural crest migrate in two pathways, and the specialization of these cells
correlates with their migratory pathways. The dorsolateral pathway gives rise to pigment cells, whereas the
ventromedial pathway forms neurons and glia of the peripheral nervous system. The question of whether the
differentiation of these cells is predetermined or dependent upon their final destination still remains. To address
this problem, we aimed to determine what genes govern the different migratory patterns and cell lineages of cells
in the trunk of the neural crest in chicken embryos. Based on a transcriptome analysis, we used in situ
hybridization and immunochemistry on candidate neural crest genes to validate the transcriptional profiles of
ventrally migrating neural crest cells. With this knowledge, the relationship between cell migration and cell lineage
in trunk neural crest cells can begin to be uncovered.
A Versatile Framework Integrating Machine Learning With PSTREAMS Streaming Data Library
Rahul Bachal
Mentor: Mani Chandy
With the increasing prevalence of the Internet of Things in our daily lives, real-time analysis of streaming data is
becoming an exciting and growing field. The PSTREAMS package enables a novice programmer to quickly and
easily apply vanilla functions to streaming data in Python without prior knowledge of the streaming architecture.
The goal of this project was to extend the PSTREAMS package to provide a framework for applying machine
learning algorithms to windows over streaming data. Steps taken to design and implement such a framework
include the development of an underlying abstraction to enable plug-and-play support for any machine learning
algorithm along with a small library of tailored incremental ML algorithms boasting increased efficiency over vanilla
batch algorithms. Enhancements such as built-in outlier and change detection were also researched and
implemented. The results show that this framework significantly reduces development time and program runtime
with the use of provided incremental algorithms. This framework has been tested extensively in the Caltech
Community Seismic Network to predict offsets between local and NTP server clock time to accurately timestamp
sensor events. The power and versatility of this framework highlights its scalability and potential use in any
scenario.
Optomechanical Modeling and Signal Processing for Analysis of Novel Implantable Interocular Pressure
Sensor to Monitor Medical Condition of Glaucoma Patients
Ashwin Balakrishna
Mentors: Hyuck Choo and Jeong Oen Lee
Glaucoma is one of the leading causes of blindness in the world and still has no cure, making it important to
develop methods to consistently and accurately measure interocular pressure to monitor the state of glaucoma
patients. The idea here is to design, build, and evaluate an optomechanical pressure sensor that is implanted into
the eye of glaucoma patients. Light is shone into the eye and based on an optical model, the air gap between the
sensor and the eye can be determined. Then the difference between the initial expected air gap and that after the
deformation of the sensor from eye pressure can be used to determine the interocular pressure using mechanical
modeling of the sensor deformation. Several sensors have been fabricated, each of which were subjected to
different pressures. Then the appropriate reflection measurements were performed. An analytical model was then
synthesized to simulate theoretical reflection data given a certain interocular pressure. Finally, a variety of signal
processing techniques were employed to compare the experimental results to theory. Further measurements and
analysis were performed with sensors implanted into a rabbit eyeball. The experimental results matched up quite
well with the theoretical predictions, demonstrating the success of the sensor.

UV Slope Measurements of 10 High Redshift Sub-Millimeter Galaxies

Ivana Barisic
Mentor: Peter Lawrence Capak
We present high-resolution Hubble Space Telescope Wide Field Camera 3 Infrared (HST WFC-3 IR) images of 10
normal galaxies at z5-6 with dust continuum and C[II] dynamical measurements from ALMA (Capak et al. 2015).
These new data enable much more accurate measurements of the rest-frame ultra-violet (UV) spectral slope ()
than were possible in the original analysis. Specifically the new data will reduce the systematic effects in
measurements due to the low signal-to-noise ground based data used in the previous analysis and allow us to
explore reddening gradients and positional offsets between the sources of the UV and FIR emission. We will also be
able to explore the link between the rest-frame UV morphology and the dynamics of the galaxies.
Time-Based Vertex Reconstruction in the Compact Muon Solenoid
Ben Bartlett
Mentors: Maria Spiropulu, Adi Bornheim, and Lindsey Gray
The Phase-II upgrades to the Large Hadron Collider will introduce a variety of new measurement devices to the
CMS, including the High-Granularity Calorimeter (HGCAL). The increase in luminosity from these LHC upgrades will
also have the undesired side effect of vastly increasing pileup to a level at which the current machine learning
vertex selection algorithms will cease to be effective. This will necessitate the development of further vertex
reconstruction (vertexing) algorithms. Using high precision timing measurements from simulated events in the
HGCAL, we design a vertexing algorithm that requires only the spatiotemporal arrival coordinates to reconstruct
the interaction vertex of a collision with sub-millimeter resolution. We also analyse how particle energy and
simulated time smearing affect this resolution and we apply this algorithm to more complex data sets, such as
H, jets, and high-pileup events. For these, we implement a set of secondary algorithms and filters to remove
poorly-reconstructed events, account for premature interactions, and resolve reconstruction ambiguity.
Numerical Simulation of Atmospheric Photochemistry in a Laminar-Flow Tube Reactor
Michael Bauer
Mentors: John H. Seinfeld and Matthew Coggon
The climatological impact and chemical characteristics of atmospheric aerosols remain very uncertain. The latest
step in this field is the experimental use of UV-lit gas-phase tube reactors to simulate days to weeks of
atmospheric photochemistry. These reactors have typically been modeled as plug-flow reactors. We have
constructed a more detailed model to describe the new laminar-flow tube reactor constructed by the Seinfeld
Group. This model assumes radial symmetry and takes into account the non-uniform velocity distribution in the
tubes flow. However, it ignores gas-phase diffusion since preliminary simulations indicated that it may not have a
large effect. A numerical simulation of this model was constructed to consider the steady-state addition of an
arbitrary number of reactants. Starting with a list of reactions and reaction rate constants, this simulation can
generate an estimate of species concentrations for each species everywhere in the reactor. It does this by
repeatedly solving an N-dimensional ODE for different reactor streamlines using backward differentiation formulae
and Newton iteration. This allows for the simulation of very large and very stiff kinetic systems within the reactor.
Experimental evaluation of the models accuracy is necessary. Diffusion may also be considered in the future.
C-H Amination of Tetrahydroisoquinoline
Katherine Bay
Mentors: Brian Stoltz and Seojung Han
Cyanocycline A is a member of the tetrahydroisoquinoline (THIQ) alkaloids, a class of natural products that display
potent biological activity as antitumor antibiotics. Isolated from the fermentation broth of Streptomyces
flavogriseus, cyanocycline A is a potent, broad-spectrum cytotoxin effective against mammalian and bacterial cells.
We envisioned a total synthesis of cyanocycline A involving late-stage CH activation as a key step. The chemistry
of CH amination offers powerful methodology for the synthesis of amine-derived materials through intramolecular
CH oxidation. Progress was made towards producing a benzo-fused oxathiazinane heterocycle from an orthosubstituted phenolic sulfamate through CH amination using a Rhodium-tetracarboxylate catalyst. The preparation
of this novel substrate contributes to the total synthesis of Cyanocycline A.
Self-Replication of One-Dimensional Colloidal Structures Through Shape Complementarity
Joseph Berleant
Mentors: Sharon Glotzer, Pablo Damasceno, and Erik Winfree
Self-replication is an essential feature of many fascinating natural phenomena, such as the replication of DNA
strands during cell division. In synthetic materials, self-replication is important in part because it is expected to be
a useful route to the scalable design of materials for complex structural assembly. Simulated self-replication of
colloidal structures has been moderately successful by relying on a set of highly specific attractive forces to control
interactions between particles; however, the ability to create such forces between particles at the nano and

colloidal scales has proven challenging. We propose and test an alternative approach to self-replication that utilizes
particle shape to confer binding specificity comparable to that achieved with multiple attractive forces. In this
scheme, a single attractive force promotes nonspecific binding between particles while shape complementarity
ensures that only desired bonds form. Self-replication is enabled through a temperature cycle that facilitates each
successive step of the replication process (Figure 1). System simulations demonstrate self-replication of a strand of
particles with some error in the replicated strand lengths. By varying properties of the temperature cycle, strands
of a particular length may be selected, making this type of scheme potentially useful in a mutation-selection
scheme for materials design.

Figure 1: A single replication cycle involves a cooling phase followed by a heating phase. During the cooling step,
monomers bind to the template strand, and at the end any adjacent monomers on a strand are bonded to create a
new strand. During the heating step, the two strands separate to serve as templates during the next replication
cycle.
Quantum Walk Frameworks and Quantum Speedup of Markov Mixing
Jeremy D. Bernstein
Mentors: Thomas Vidick and Stacey Jeffery
Algorithms based on random walks have a wide applicability: from problems in theoretical Computer Science to
Google's PageRank algorithm. We investigate two frameworks for quantising algorithms on reversible Markov
chains that confer polynomial speedup over the classical counterparts. The MNRS framework is based on the
spectral properties of Markov chains, and the Belovs framework is based on the electric network interpretation.
In an effort to unify the frameworks, we rediscover a quantum algorithm for sampling from the stationary
distributions of ergodic, reversible Markov chains due to Wocjan and Abeyesinghe. The algorithm makes use of
phase estimation on the Szegedy walk operator to implement a reflection in the stationary distribution la MNRS.
The complexity is asymptotically better in the size of the spectral gap than both the classical power method and a
quantum algorithm using adiabatic state generation due to Aharonov and Ta-Shma. We use this algorithm to
compare the power of the Belovs and MNRS frameworks.
High Strain-Rate Dynamic Behavior of Magnesium and AZ31B
Moriah Bischann
Mentors: Guruswami Ravichandran and Owen T. Kingstedt
Magnesiums implementation as a light weight structural material has lagged because its hexagonal-close-packed
(HCP) crystal structure leads to complex deformation response. Within this work we examine the high strain-rate
(102 to 103 s-1) dynamic response using the split-Hopkinson pressure bar. Experiments were conducted using
specialized striker bars to promote a constant strain-rate, strain-rate increment, or strain-rate decrement during a
single loading event. Thermo-mechanically processed magnesium (Mg) and Mg alloy AZ31B were examined along
various loading directions to explore the strain-rate sensitivity of the different deformation mechanisms, such as
deformation twinning or basal slip, which occur under a prescribed loading condition. The results of these
experiments capture the anisotropic material response of Mg and AZ31B with respect to loading orientation.
Magnetic Field Profile of a Third-Scale Model for the nEDM Experiment
Aritra Biswas
Mentors: Bradley Filippone and Simon Slutsky
Discovery of an electric dipole moment in neutrons (nEDM) would be a novel instance of CP violation, with
implications for extending the Standard Model and potentially helping explain matter-antimatter asymmetry.
Experiments using shifts in polarized neutron spin-precession frequency to measure the nEDM are prone to a
geometric phase (GP) effect, caused by gradients of the magnetic field, that can create a false signal. Preventing
the GP effect requires precise engineering to create a space-uniform magnetic field. We present a third-scale
prototype of a shielded magnet suitable for a more precise nEDM measurement, with improvements over earlier
models. The field is produced by a cos coil wound with superconducting (SC) wire. Two cylindrical shields made

of ferromagnetic Metglas and SC lead surround the magnet; the lead shield is closed on top and bottom with SC
lead endcaps. An aluminum shell surrounds these components and serves as a vacuum chamber, cooling its
interior to 4 K such that the coil wire and lead shield become SC. A cavity in this shell serves as a warm bore,
allowing a magnetic probe to explore the field around fiducial volumes which will be used to measure the nEDM in
the full-scale experiment. A magnetic field profile of components of this prototype is presented.
Exploring the Reactivity of Diazo Compounds: Discovery of a Novel Rhodium-Catalysed
Cyclopropenation of Diazooxindole and Identification of an Exceptionally Simple System for the
Oxidation of Aryl Diazoacetates to -Ketoesters
Peter Bolgar
Mentors: Brian M. Stoltz and Nicholas OConnor
A novel, highly strained, cyclopropene-containing spirooxindole has been synthesised by the insertion of a rhodium
carbenoid derived from N-methyl-3-diazooxindole into the -bond of a terminal alkyne. This reaction provides a
new entry into the spirooxindole class of molecules, which often have important biological functions. We also
present a systematic study of the oxidation of electron rich aryl diazoacetates to -ketoesters by warming in dry
DMSO. The use of cheap and environmentally friendly DMSO provides a major advantage over methods using
expensive and toxic metal catalysts. Furthermore, the mild nature of DMSO makes this method compatible with a
variety of common functional groups susceptible to oxidation under other conditions.

Three-Dimensional Microstructures as Effectively Transparent Front Contacts for Solar Cells

Aleca Borsuk
Mentors: Harry Atwater and Rebecca Saive
For solar cells to achieve optimal photovoltaic performance, front contactselectrodes at the surface of the solar
cellmust offer excellent electrical conductance and optical transmission. At present, however, front contacts do
not deliver on all criteriae.g. metal grids shadow otherwise active surfaces, while transparent conducting oxides
absorb photons. In the present work, a novel front contact architecture is realizednamely, three-dimensional
(3-D) microstructured fingers that are highly conductive and effectively transparent. Here, fine metal grids with 3.0
m finger width and 40 m periodicity were first defined lithographically. These were superimposed with pyramidal
structures with >45 degree angles, which re-direct incident light to the active surface of the solar cell. Pyramids
were written with the Nanoscribe direct laser writing system, in which high intensity laser pulses induce
polymerization of photoresist at nanoscale volumes. To achieve efficient re-direction of light, 3-D surfaces were
selectively coated with a metal film using directional deposition. Spatially resolved laser beam induced current
measurements (LBIC) show that 3-D contact patterning leads to significantly improved photocurrent as compared
with planar contacts. Current-voltage (IV) curve characterization of 3-D contacts additionally indicates enhanced
absolute current and increased efficiencies in state-of-the-art solar cells.
Using Textured Silicon as an Antireflective Treatment for Microwave Lenses
Arjun Bose
Mentors: Sunil Golwala and Simon Radford
Since modern astrophysics requires much larger millimeter and submillimeter detector arrays than ever before,
previous antireflection (AR) treatments, such as grooved designs and laminated plastic coatings, are begin
presenting deficiencies. These deficiencies include distortions from inability to effectively vary the index of
refraction or deformities from thermal contraction and expansion, and noise from the absorbing and reemitting
light. An option with fewer drawbacks may be to create an antireflection treatment with layers of silicon with
various holed patterns in order to create a gradient index (GRIN) coating, which would solve the previously
mentioned issues for the millimeter and submillimeter spectrum.
Engineering Prolyl tRNA Synthetase for Accommodation of Non-Canonical Amino Acids
Mary Boyajian
Mentors: David Tirrell, Katherine Yan Fang, and Seth Lieblich
Current methods of engineering recombinant proteins primarily involve the replacement of one canonical amino
acid with another; however, for certain proteins of heavy research and commercial interest, all reasonable
canonical mutations have been extensively explored. Diabetics is a disease that affects millions in the world; it is
the result of an inability to produce or respond to insulin. As a result, insulin has been the subject of intensive
engineering efforts for decades. Few, if any further, engineering opportunities remain undiscovered using canonical
amino acids; however, the replacement of canonical amino acids (cAAs) with non-canonical amino acids (ncAAs) is
relatively new and opens new unexplored chemical space wherein beneficial engineering via mutations is possible.

Previous research in the field has demonstrated that hundreds of ncAAs can be incorporated into newly synthesized
proteins by bacterial and mammalian cells. These ncAAs have unique side chains and can introduce novel chemical,
physical, and biological changes that cannot be achieved with the natural cAAs. Aminoacyl-tRNA synthetases
(aaRSs) are responsible for catalyzing the specific ligation of amino acids to their cognate tRNAs; in most cells,
there are at least 20 aaRSs, one for each natural amino acid. While some ncAAs are promiscuously charged to
tRNA by aaRSs, most ncAAs require a mutant aaRS to enable protein translation. Insulin has a single proline in its
mature form: B28; this position is known to be critical for controlling insulin oligomer states and has served as the
most important site of alteration and modification in new biologic drugs. In this project, certain mutations will be
made to the prolyl tRNA synthetase (ProRS) to accommodate the different side chains on ncAAs in order to explore
new protein sequence space for the engineering of improved insulin variants.
Spectroscopic Quantification of Structurally-Bound Water in Pyroxenes and Olivines, With Applications
to Meteorites
Alec Brenner
Mentor: George Rossman
I present polarized near-infrared absorption spectra of several mineral species extracted from meteorites. I place
emphasis on absorption features associated with O-H stretching (peaks near wavenumbers of ~3500-3300 cm-1),
which I use as proxies for the presence of structurally-bound water (or OH-). The meteorites treated in this analysis
include the Fukang pallasite (a stony-iron cumulate originating from the core-mantle boundary of a protoplanet),
the Esquel pallasite (of similar origin), and ALH84001 (a shergottite of Martian crustal igneous origin). In the cases
of both pallasites, the mineral species I analyze for O-H content is forsterite (Mg2SiO4, the magnesic olivine), and I
compare these spectra to existing IR spectra of terrestrial forsterite. For ALH84001, the analytical material is the
orthopyroxene cumulate that dominates the meteorites composition (~97 vol%). Additionally, I compare the O-H
region spectra of the ALH84001 orthopyroxene to existing spectra of terrestrial enstatite (Mg2Si2O6, the magnesic
and most common orthopyroxene), as well as new IR spectra of wollastonite (CaSiO3, a calcic pyroxenoid) and
spodumene (LiAl(SiO3)2, the lithic clinopyroxene).
Non-Linear Entropy Inequalities for 2D Conformal Field Theories
Emory Brown
Mentors: John Preskill and Ning Bao
Significant work has gone into determining the minimal set of entropy inequalities to determine the holographic
entropy cone. It was previously shown that there must exist at least one non-linear entropy inequality when
describing states on a 2D conformal field theory. We report on five sets of these non-linear inequalities. We also
present an equality obtained by application of a hyperbolic extension of Ptolemy's theorem to a 2D conformal field
theory.
Single Molecule Investigation of Phage-Mediated Gene Transfer in E. coli
Robin Brown
Mentors: Rob Phillips and Griffin Chure
With an estimated global population of 1031, bacteriophages (viruses that infect bacteria) are the most numerous
biological entities. Despite their abundance, much about their basic biology as well as their influence on microbial
communities remains unknown. In a process called transduction, a bacteriophage accidentally packs segments of
the host bacterias genome instead of viral DNA in its capsid and donates it to the recipient. The Phillips lab has
previously designed and implemented a system using recognition sites and fluorophores to visualize the spatial
distribution and movement of DNA within the cell. We modified this system to track phage infection with the
intention of comparing transduction frequency to infection frequency. Due to the low frequency of transduction,
this system seems to be impractical for the study of transduction. However, I was able to observed infected cells
lyse and infect neighboring cells. As a result, this system seems to work well to track phage infection and
propagation through populations of E. coli. For example, we will be able to time how long it takes an infected cell to
lyse, and then how long it takes neighboring cells to become infected. Furthermore, we can the range of phage
infection from the location of the lysed cell.
Modeling Turbulent Buoyant Flows From Direct Numerical Simulation Data Sets
Nicholas A. Buoniconti IV
Mentors: Guillaume Blanquart and Julien Claret
Turbulent Buoyant Flows (TBF) occur when discontinuities are generated in a fluids flow regime due to a variance
in the densities of fluid particles in that flow. TBFs are present in a multitude of natural and engineering
phenomenon. Understanding these flows is a fundamental problem that we have not yet solved. Herein we
evaluate a variety of data sets developed from Direct Numerical Simulations (DNS) with a Matlab code framework.
The findings include a relation for the dependence of velocity on density, a model of the effect of partial buoyancy
on turbulent flow behavior, and an understanding of the effect of energy dissipation rates on flow velocities. The
results supported existing theories for turbulent behaviors found in the literature. Additionally the different

components contributing to the flows turbulence were successfully separated and modeled. In the future this
research can be utilized to ease the computing cost of TBF simulations. Additionally the insight gained from the
results will be important in supporting theories of the TBF behavioral framework and continuing the understand of
the effects of density variance on TBF.
Investigating a Potential Role for Karyopherin- in the Assembly of the Nuclear Pore Complex
Nina Butkovich
Mentors: Andr Hoelz and George Mobbs
The nuclear pore complex (NPC) is an essential, massive macromolecular structure composed of approximately 30
different proteins known as nucleoporins, providing the sole gateway between the nucleus and cytoplasm in
eukaryotic cells. While significant regions of its architecture have recently been resolved, the NPCs assembly is not
well characterized. Karyopherin-, a nuclear transport factor, has been shown to interact with several nucleoporins
and is hypothesized to play a role in the assembly and oligomerization of the NPCs distinctive ring like structures.
To further study its role in assembly, we have investigated how Karyopherin- interacts with Nup53, a flexible
linker which interacts with a wide array of nucleoporins, and Nup133, the base of the coat nucleoporin complex. We
have successfully expressed, purified and crystallized Karyopherin- with a short peptide corresponding to the
minimal binding region of Nup53. We have also identified a minimal Karyopherin- binding region in Nup133
through testing various truncated constructs using size-exclusion chromatography. This work shows that
Karyopherin- forms physiologically significant interactions with two crucial nucleoporins, which opens the door for
a larger role in NPC assembly.
Designing an in vitro Translation System and Engineering Methane Monooxygenases
Slava Butkovich
Mentors: Steve Mayo and Alex Nisthal
Molecular engineering is a powerful tool that can be used to improve many aspects of life. In this research, two
separate projects utilizing molecular engineering are pursued. One project involves creating an in vitro translation
system, using RNA to synthesize protein instead of starting with DNA. By modifying tRNA synthetases in this
system, tRNA could be made to accept different amino acids than usual; a successful system could be used to
better incorporate unnatural amino acids in proteins. Current work strives to show that this theoretical system can
be implemented. Another project involves discovering methods of more efficiently turning methane into methanol.
Methane is readily available but relatively difficult to transport, while methanol is easier to transport. Thus,
expanded use of this conversion could reduce waste and help to prevent the buildup of carbon dioxide in the
atmosphere. Methane monooxygenases are proteins naturally found in methanotrophic bacteria that can perform
this conversion under ambient conditions. Current work involves engineering methane monooxygenases and
developing a folded protein screen in bacteria and yeast. Further research will be able to better implement the
proposed in vitro translation system and to improve engineered methane monooxygenases for more efficient
conversion of methane to methanol.
Structural and Functional Characterization of the Interaction of mRNA Export Factor Gle1 With
Cytoplasmic Filament Protein yNup42/hNupl2 at the NPC
Sarah Cai
Mentors: Andr Hoelz and Daniel Lin
The mRNA export factor Gle1 is an essential nucleoporin that functions in the export of poly(A)+ mRNA through the
nuclear pore complex (NPC). Mutations in Gle1 can result in defects in mRNA export amyotrophic lateral sclerosis
(ALS) is linked to a mutation removing the Nupl2 binding domain on Gle1. Gle1 has two known binding partners,
one of which is the cytoplasmic filament protein hNupl2 (yNup42 in S. cerevisiae). A minimal binding construct was
determined, expressed in E. coli, and purified. We solved the atomic structure of the yGle1-yNup42 complex,
refined to 1.75 resolution. The structure reveals several key residues that mediate the interaction, and
experiments are being done in vivo with S. cerevisiae using mutants to validate the structure and to study NPC
localization patterns of yGle1 and yNup42. The homologous human complex (hGle1-hNupl2) has been purified and
is in the process of screening for crystallization conditions. Understanding the atomic structure of these interactions
can be applied toward the study of diseases such as ALS that result from mRNA export defects.
Increasing the Resolution of N-Body Cosmological Simulations
Chris Cannella
Mentors: Andrew Benson and George Djorgovski
A method for inserting Monte-Carlo derived branches into the dark matter halo merger trees produced by N-Body
simulations (such as those from the Millennium Run) is created and tested. Such a method allows the artificial
increase in the resolution of merger trees generated by N-Body simulations, allowing accurate modelling of galactic
formation and evolution based on N-Body simulations.

10

The Effects of Accretion on the Age Spreads of Young Stars

Lyra Cao
Mentor: Lynne Hillenbrand
Stars are thought to form from an initial period of hot spherical accretion that accounts for most of the mass,
followed by cold disk accretion that occurs throughout a significant portion of the stars early evolution. Since
accretion histories are unknown for individual stars, the effect of the accretion is to produce an error on stellar
parameters, including luminosity, radius, and age. Recent studies have shown that these errors due to cold
accretion are systematic and large for low-mass stars, producing older ages and smaller radii than expected
(Baraffe et al. 2009). However, a detailed study of accretion on higher mass stars has not been done. Theoretical
and observational arguments suggest that rapid accretion, suitable for high-mass stars, is likely to be hot rather
than cold (Hartmann et al. 2011). To treat high-mass star formation correctly, we develop routines to perform
episodic hot accretion in the Modules for Experiments in Stellar Astrophysics (MESA). We present the correct
treatment of protostellar accretion, with plausible accretion histories, on stars ranging from 0.1 to 8 solar masses.
Toward the Synthesis of a High-Valent Terminal Mn(IV)-Oxo Motif Supported by a Tripodal
Trimanganese Framework as a Model for the Oxygen-Evolving Complex of Photosystem II
Kurtis Carsch
Mentors: Theodor Agapie and Graham de Ruiter
A series of homometallic, tetramanganese clusters of structures [LMn3(PhPz)3OMn][OTf]x (x = 0, 1, 2; PhPz = 3phenylpyrazolate) and [LMn3(F2PhPz)3OMn][OTf]x (x = 1, 2; F2PhPz = 3-(2,6-difluorophenyl)-pyrazolate) were
synthesized to provide insight into the OO bond coupling mechanism accomplished by the oxygen-evolving
complex (OEC) of photosystem II. The rational, stepwise synthesis of [LMn3(PhPz)3OMn][OTf]2 (3) was
accomplished starting from precursor LMn3(OAc)3 (1). Cyclic voltammetry of 3 revealed two reversible redox waves
corresponding to [MnII4]0/[MnII3MnIII]1+ (-1.66 V) and [MnII3MnIII]1+/[MnII2MnIII2]2+ (-0.66 V) oxidation and reduction
events. [LMn3(PhPz)3OMn][OTf] (4) was obtained upon reduction of 3 via cobaltocene (CoCp2). Both 3 and 4 are
further characterized by single-crystal X-ray diffraction, supporting the respective oxidation state assignments as
[MnII2MnIII2]2+ and [MnII3MnIII]1+. Treatment of 3 and 4 with oxygen-atom transfer reagents iodosobenzene (PhIO)
and TBAIO4 suggest intramolecular CH hydroxylation of the adjacent phenyl motif to yield
[LMn3(PhPz)2(OPhPz)OMn][OTf]2 (5). A similar reaction with [LMn3(F2PhPz)3OMn][OTf]2 (7) and PhIO or TBAIO4
suggests CF activation of the difluorophenyl pyrazolate motif, accompanied by coordination of F- onto unreacted
starting material with subsequent one-electron disproportionation to yield [LMn3(F2PhPz)2(OFPhPz)OMn][OTf]2 (8)
and [LMn3(F2PhPz)3OMnF][OTf]2 (9). Both CH and CF activation via addition of PhIO are examined via
independent synthesis routes employing substituted pyrazolate scaffolds as well as addition of XeF2 for 9.
Computational studies based on density functional theory (DFT) are provided to address the observed reactivity
with PhIO. Efforts to access more electron-deficient configurations relevant to the OEC are further discussed.
Investigating the Role of Magnetite in Cryopreservation
Thomas Chaffee
Mentor: Joseph Kirschvink
Recent technological advances have utilized alternating magnetic and electric fields in freezers that are able to
supercool food and keep it almost perfectly preserved. The mechanism by which this supercooling is enabled has
not yet been discovered, but there are indications that biologically precipitated magnetite plays a key role in ice
crystal nucleation. Our proposed mechanism is that the alternating magnetic field causes magnetite crystals in the
food to oscillate, preventing damaging ice crystals from forming, keeping the cells better preserved than in
traditional freezing. Our goal is to investigate the magnetite content of fruits and vegetables, to determine the
importance of magnetite crystal content to this freezing technology.
Study of the CMS ECAL Calibration and Timing Performance With
Kai Chang
Mentors: Maria Spiropulu, Adi Bornheim, and Emanuele Di Marco

0 Decays at 13 TeV

The CMS ECAL is capable measuring the energy of high energy photons with a precision of about 1% and their time
of arrival in the detector with a precision of around 100 ps. The calibration of the detector in situ with neutral pion
decays into two photons is a crucial step to achieve the energy resolution performance. To date the timing
response is calibrated with generic calorimeter clusters which have the characteristics of electromagnetic
interactions. One important systematic effect limiting the timing performance is the radiation induced transparency
change of the crystals. This modifies the optical path of the scintillation photons in the crystals which changes the
time response. In this project the measurement of the crystal transparency will be used to study this dependency
an derive a correction for the timing measurement. The
a clean benchmark sample to test the corrections.

0 calibration sample and analysis framework will serve as

11

Engineering Nucleic Acid Probes for Delivery, Localization, and Branch Migration in Mammalian
Cell Culture
Ann Chen
Mentors: Niles Pierce and Mikhail Hanewich-Hollatz
Because of the predictability of Watson-Crick binding thermodynamics in nucleic acids and the potential to interface
with existing nucleic acid pathways, nucleic acid nanotechnology has the potential to be a powerful tool for the
development of novel techniques in biology. The engineering of nucleic acid mechanisms, a biochemical interaction
of nucleic acid species that produces a unique output, allows for a range of programmable behaviors. Such
engineered systems are ideal for conditional gene regulation - the control of the expression of genes either in a test
tube or within a living cell - because they can generate biologically functioning species conditionally through
interactions with endogenous nucleic acid species. In this study, we characterize nucleic acid mechanisms,
designed using NUPACK (Nucleic Acid Package), that have inputs of arbitrary sequence and are designed to
produce biologically functional outputs in mammalian cells. To validate these mechanisms, microinjection assays
are used to study localization of fluorescently-labeled species. We deliver a pre-defined volume of prepared
solution to HEK 293 cells at a relevant working condition. After injection, GFP and DsRed expressions are monitored
to determine cellular behavior. We provide a characterization of GFP knockdown by the conditionally generated
output upon microinjection of small conditional RNAs (scRNAs) into HEK 293 cells containing the target DsRed
input. After validation of a particular nucleic acid mechanism via microinjection assays, additional mechanisms can
be designed and characterized.
Establishing Methods of Transgenesis in the Nematode Bursaphelenchus okinawaensis
Anthony Chen
Mentors: Paul Sternberg and Ryoji Shinya
The fungal-feeding hermaphroditic nematode Bursaphelenchus okinawaensis is an emerging laboratory model
useful for understanding the characteristics and disease mechanisms of various types of parasitic nematodes.
Recently, Dr. Shinya and I developed a forward genetics system in B. okinawaensis and successfully generated a
large number of mutants. Therefore, the establishment of reverse genetics and transgenic tools are needed to
accelerate the genetic study of B. okinawaensis. This research attempts to establish a reliable method of
transgenesis for the nematode B. okinawaensis by using microinjection, microparticle bombardment, and CRISPRCas9, all of which have been used extensively in C. elegans study as well as in other organisms. Establishing a
method of transgenesis would allow for deeper understanding of the genetic processes present in the nematode.
Because some groups of parasitic nematodes currently negatively affecting agriculture and public health are very
similar to B. okinawaensis, this new information about B. okinawaensis should also be very useful in developing
genetic methods for them.
Characterization, Simulation, and Optimization for III-V Solar Cell Contact Grids
Kevin C. Chen
Mentors: Harry A. Atwater and Cristofer A. Flowers
High material quality, high performance III-V solar cells are particularly attractive solar energy converters for
concentrating and spectrum splitting photovoltaic systems. To achieve high efficiency power production, it is
especially important to optimize the metal contact grid at the front of the cell. The goal is to minimize power losses
by finding a balance between resistive loss in conduction and amount of light reflected away from the cell on sparse
metal contact features. The circular transfer length method was employed to extract sheet resistance and contact
resistivity of III-V materials. These data enable accurate prediction of cell electrical performance using HSPICE
simulations, and genetic algorithmic approach will offer insights on the ideal geometry of the front metal contact
for maximum power conversion efficiency.
Data-Matching Algorithms for Fabry-Perot Intraocular Pressure Sensors
Oliver W. Chen
Mentors: Hyuck Choo and Jeong Oen Lee
Glaucoma is a leading cause of blindness, affecting an estimated 4 million Americans and 70 million individuals
globally. The Choo lab has developed a highly miniaturized IOP monitoring system using a nanophotonics-based
implantable IOP sensor with remote optical readout that can be adapted for both patient and research use.
However, data collection of IOP using such a micro-sensor results in a burdensome amount of background spectra
mixed in with a few useful sensor-generated spectra. We created a data extraction algorithm based on the optical
properties of the Fabry-Perot cavity of the sensor to separate the background spectra from the sensor spectra.
Subsequently, we generated a data-matching algorithm to choose the best fitting theoretical spectra for each
experimental spectrum. With these two algorithms, we calculated the IOP exerted on the sensor in optical test
bench, ex vivo, and in vivo settings.

12

Engineering Protein Matrices for Production of -Like Cells for Treatment of Type 1 Diabetes
Desnor Chigumba
Mentors: David A. Tirrell and Mark T. Kozlowski
Type 1 diabetes is a chronic condition in which patients cannot produce their own insulin. Previous studies have
shown that it is possible to manipulate post-natal progenitor cells to differentiate directly into -cells for
transplantation by using laminin and methylcellulose to provide the appropriate biochemical and mechanical cues
respectively. By using engineered protein hydrogels with an encoded laminin sequence and leucine zippers, we can
make a material that provides both cues. As the cell culture is 3-dimensional, this would enable us to suspend the
cells in solution. Our primary objective was to produce a hydrogel which would be soluble at 37C. Previous protein
constructs were not soluble and we suspect this was because they had low LCSTs. Therefore, we altered the
sequence to increase solubility. We cloned a PC10-lam-C10P protein where the C domain is hydrophilic and does not
have an LCST. PE6-lam-E6P, PE6-trunc lam-E6P and PE6-lam-E6P C3A were also constructed as modifications of a
previous elastin PEP construct. All constructs were soluble and MALDI-TOF MS suggests we obtained our protein of
interest. All constructs are sent to City of Hope Hospital for cell culture studies.
Developing Nickel Inverse Opal Cathodes for Lithium Air Batteries
Evelyn S. Chin
Mentors: Julia R. Greer and Dylan D. Tozier
Lithium air batteries have an energy density up to 10 times greater than that of lithium ion batteries. This property
makes Li-air battery systems very desirable, however there are many challenges to making these systems. Recent
research has focused on the use of carbon based cathodes because carbon provides a high surface area ideal for
maximizing battery chemistry. However, these cathodes lead to unwanted byproducts which hinder the ability to
recharge these batteries. Metals with nanoporous structures can be used to prevent production of these byproducts
as they can both catalyze the reaction and provide the desired high surface area. In this study, nickel inverse opal
structures were synthesized by electroplating nickel through self-assembled polystyrene opal structures. The
inverse opal structures were then tested in a molten salt battery system.
Tools for Developing Apps for the Internet of Things
Wonwoo Cho
Mentors: K. Mani Chandy and Julian Bunn
Basic part of Internet of Things (IoT) includes sensors, processors, and codes. We develop a framework that could
be applied to IoT, and it should be easily handled by novice users. The framework is based on several sensors,
Arduino, Raspberry Pi, and python codes running on Raspberry Pi. One of popular programming languages for
developing IoT applications is python. Using python, novice programmers write a standard non-stream codes. For
wrapping the non-stream function codes to generate streams, we used a python package called PStreams which
enables working with streams. Testing with several kinds of sensors and connection types, we developed a basic
example platform that can read sensor data, and generate data streams for application. We can expect novice
users and programmers can develop their own useful streaming system based on our platform. Also it could be
applied to further related area like Machine learning, and visual UI.
Numerical Analysis of Thermocapillary Patterning of Polymer Films
Dahan Choi
Mentors: Sandra M. Troian and Chengzhe Zhou
Thermocapillary forces are believed to be a dominant cause of spontaneous growth of nanopillars from molten
polymer nanofilms subject to a transverse thermal gradient. According to this mechanism, the difference in
polymer film thickness causes disturbances in film surface temperature, which results in the rearrangement of film
material towards colder regions. We examine the highly nonlinear equation governing the evolution of the film
height under thermocapillary forces. Linear stability analysis was used to model a flat film perturbed by a small
amount, and to calculate dispersion relation between the growth rate and the wavenumber of the perturbation.
Several different initial film configurations were numerically simulated. Results yielded a nearly perfect match
between LSA and numerical computations for early times, but failed to capture the long term behavior of the
instability. Subsequent work will be aimed at understanding this long term behavior of the instability.
Localization of Photoactive Rubpy Label by Strategic Labeling of CssII Scorpion Toxin
Nelson Chou
Mentors: Robert Grubbs and Melanie Pribisco
Visual impairment due to incurable retinal degenerative diseases such as retinal pigmentosa and age-related
macular degeneration is characterized by the loss and damage of photoreceptors, which in turn leads to
degeneration of the visual field. However, vision can be restored by bypassing damaged photoreceptors and
activating surviving retinal ganglion cells. Previous studies have shown RubpyC17 inserts evenly into a cell
membrane. Upon illumination, Rubpy is excited to Rubpy* which can reduce or oxidize the redox active molecules

13

around the ion-gated channel, causing a buildup of negative charge leading to action potential firing and
transmittance of light information to the brain. Such method, though useful, has no inherent targeting, therefore
may require higher doses of toxic ruthenium than necessary. We hope to localize this phenomenon by tagging the
Rubpy label to CssII, a 66-amino acid peptide that binds with nanomolar affinity to Nav 1.6 ion-gated channels of
concentrated near ganglion cells. To form full-length CssII, three fragments must be synthesized and joined by
native chemical ligation. We propose to label the native His50 residue of CssII with Rubpy. If the binding affinity is
retained, the Rubpy label will be localized and oriented such that, upon illumination, localization of accumulation of
negative charge so that action potential firing can be induced efficiently. Recently, we also identified a new 31
amino acid fragment that contains important binding sites of CssII as well as the His50 residue and may provide a
more efficient route to targeted delivery of the molecular artificial retina.
Machine-Readable Protocols and Rapid-Prototyping for Synthetic Biology Research
Ahammad Rafsan Chowdhury
Mentors: Richard M. Murray, Scott C. Livingston, and Sean R. Sanchez
Natural biological systems have always been challenging to manipulate. In recent years advancement has been
made in synthetic biology to encourage an engineering approach. A central concern of synthetic biology is
predictable engineering of chemical processes within cells, which requires repeatable experiments. The aim of this
project is to develop a machine-readable format for protocols. Focus is placed on the construction of liquid handling
robots that automate parts of a protocol in a wet lab. A prototype robot made of acrylic plastic has been devised.
The robot is designed in SOLIDWORKS. Tools such as laser cutter, water jet, lathe and milling machine are used to
construct the robot. The robot is capable of movement in three-dimensional axis using stepper motors and timing
belts. Work needs to be undertaken to create the mechanism that can be attached to the robot with the capability
to pipette liquid.
Designing a Thermal Management System for Electric Vehicle Performance
Edward Chu
Mentor: Guillaume Blanquart
My project seeks to address the issue of thermal management and cooling of the various components of electric
vehicles, or EVs. This is an important issue; although electric vehicles have come quite a long way in the past
decade, with corporations such as Tesla leading the way, they are not quite up to par with conventional vehicles in
terms of both performance and cost. Range, refueling time, and cost are simply not up to par with consumer
expectations. This is partly due to suboptimal temperature management of various EV components. My project
seeks to improve the cooling system of EVs with a potentially novel design. We will use a variety of methods and
materials to keep an electric vehicles components each running at the temperature for maximum efficiency. To
achieve this end, we first started with some thermal analysis simulations in SolidWorks, to model how the heat
would flow and distribute in models of the battery pack, motor, and motor controller. Using that data, and data
gathered from actual experimentation on the CSVC electric vehicle, an optimal design will be apparent. Improving
the thermal management system of electric vehicles will make them much more viable in the consumer market,
helping to battle against pollution and climate change.
Nanoindentation Characterization of Acrylic-Zinc Oxide Systems as Model Artists Paints
Ya Lun Chuan
Mentors: Katherine T. Faber and Matthew T. Johnson
In order to better preserve paintings that are important to our cultural heritage, conservation scientists are looking
to characterize paint systems. Because it takes up to decades for paint to cure, it is important for conservation
scientists and conservators to understand how these paint systems change mechanically as they cure. In this
project, model paints of acrylic binders and zinc oxide pigments are produced and characterized as a function of
pigment concentration with acrylic systems were compared to a fast drying alkyd binder. As the paints were set to
cure in a dark, dry environment at room temperature, increases in the hardness and elastic modulus were
observed for all pigment fractions. Scanning electron microscope images show that the pigment particle sizes are
small relative to the indentation impression, proving that the indentation data are sampling both binder and
pigments. Similar future experiments can be conducted on different binders, pigments, and pigment percentages.
Beltrami Differentials and the Gauss-Bonnet Theorem for Noncommutative Two Tori
John Michael Clark
Mentor: Farzad Fathizadeh
The Gauss-Bonnet theorem for surfaces serves as the prime example of a novel connection between geometry and
topology. There is a spectral formulation of the theorem that has stimulated the investigation of its analog in
noncommutative geometry. The theorem has been proved for translation-invariant conformal structures on the
noncommutative two torus (T2) that are associated with arbitrary complex numbers in the Poincar half-plane. We
study the analog of the Gauss-Bonnet theorem and develop an understanding of the local differential geometry of
T2 equipped with metrics that belong to general conformal structures introduced by Beltrami differentials. We

14

introduce a complex structure on T2 and go on to compute its scalar curvature using the noncommutative
pseudodifferential calculus to follow the heat kernel method for computing the heat coefficients of elliptic
differential operators. We then investigate whether the scalar curvature vanishes under the volume form, namely
whether the analog of the Gauss-Bonnet theorem holds.
Focused Ultrasound Induces Currents Across the Cell Membrane of Xenopus laevis Oocytes
Jonathan R. Clements
Mentors: Mikhail Shapiro and Jrme Lacroix
It has been shown that focused ultrasound (FUS) can induce neuromodulation in the brain. In my experiments I
tested the hypothesis that this phenomenon is caused by the opening of mechanosensitive ion channels
(MS channels) in neurons. This would be caused by the mechanical pressure of FUS creating a membrane stretch
which opens the channels. To model a neuron, we used Xenopus laevis oocytes, which express various endogenous
MS channels. We used a two-electrode voltage clamp to detect any currents generated across the membrane of the
oocytes and subjected them to FUS. At first, we were unable to induce a current, so we attempted to increase the
effect of FUS on the membrane by subjecting the oocytes to various treatments, but instead found that,
interestingly, only a narrow range of FUS frequencies are sufficient to induce a current, which can at times be as
large as 2 A. The nature of the current and its mechanism are still under investigation.
Incorporating Realistic Feedback From Super Massive Black Holes in Simulations of Galaxy Formation
Matthew J. Colbrook
Mentors: Phil Hopkins and Xiangcheng Ma
Supermassive black holes (SMBHs) are found at the centre of massive galaxies. One of the ways these SMBHs
interact with their host galaxies is through energy and momentum delivered through relativistic, high energy
particle jets. This may give rise to many interesting observed relations between black hole and galaxy properties,
and may also play an important role in star formation within massive galaxies. To model feedback better, we need
to better understand some fundamental processes that are not well-studied, among which is the problem of metal
diffusion, important in galactic chemical evolution, cooling etc. The interstellar medium is mostly supersonically
turbulent and current literature sub-grid models of diffusion are generally over-simplified, producing unrealistic
results in cosmological simulations. We ran numerical simulations in GIZMO, a magneto-hydrodynamics and gravity
code on a periodic turbulent box in two and three dimensions. We found that the tracer was well modelled by a
diffusion type equation with a diffusivity dependent on the wave number in Fourier space. Physically this
corresponds to eddies of different length scales. Scalings were found to agree well with Burgers model of shock
driven turbulence. Finally, in the coming weeks we hope to extend this analysis to a galactic disk where shearing
becomes important dynamically.
Implicit Bias and Social Decision Making in Autism Spectrum Disorder
Caitlin Cooper
Mentors: Ralph Adolphs and Damian Alexander Stanley
Trust is indispensable to forming relationships, and is influenced by implicit social biases of which we are often
unaware, such as racial bias. Individuals with Autism Spectrum Disorder (ASD) are known to have atypical social
processing, however, previous work from our lab shows that their implicit social biases remain intact. We
investigated whether implicit social biases in ASD maintain their influence over social behavior. Data from 200
neurotypical participants was collected online using MTurk. Each participant completed two economic games, where
they viewed faces of many different partners and made economic decisions that reflect trust and altruism, and a
task in which they explicitly rated the trustworthiness of many faces. They then completed an implicit association
test (IAT) and finally various questionnaires designed to measure traits associated with ASD as well as explicit race
bias. Consistent with previous findings, we found individuals race disparity in trustworthiness ratings correlated
with their implicit race bias. Interestingly this was not the case for the economic games. We are in the process of
collecting data from ASD participants and matched controls as well as probing whether ASD traits in the general
population account for variation in the association between trust and implicit race bias.
Wing Threats Are Sufficient to Establish Dominance in Drosophila melanogaster
Alberto Corona
Mentors: David J. Anderson and Brian Duistermars
Animals from flies to humans express extremely complex aggressive social behaviors that have been studied for
many years. However, aggressive behaviors remain poorly understood. It is essential to understand the neural
circuitry controlling innate behaviors, such as aggression, to further investigate psychiatric disorders in humans.
We have recently isolated a small set of neurons that when activated or silenced, promote or abolish, respectively,
a single aggressive behavior in Drosophila, the wing threat. During an aggressive encounter, a wing threat consists
of raising one or two wings upward at a 45-90. In this project, we investigated whether wing threats are sufficient

15

for dominance formation using artificial activation of identified wing threat cells. To study this behavior, and despite
its variability, we utilized a semi-automated classifier to accurately score wing threats contained in large datasets.
Measuring the relative level of aggression between a pair of flies and relative territoriality, we conclude that wing
threats are sufficient to establish dominance in conspecific interactions in Drosophila melanogaster.
Nondestructive Mapping of Hybrid Rocket Fuel Grains
Adrian Costantino
Mentors: Brian Cantwell and Sergio Pellegrino
The regression rates of solid fuels used in hybrid rocket motors are a critical parameter in the design and
development of hybrid propulsion systems. Currently, it is difficult to measure fuel geometries accurately, thus
causing difficulties in accurately describing regression rates, especially for smaller scale systems. Nondestructive
techniques can solve this problem, as it is not necessary to get tooling into the ports. An apparatus that maps the
fuel grain employing an immersed ultrasound transducer has been manufactured; its results will be the primary
focus of this study. The full geometry of the fuel grains can be generated pre and post fire, which may lead to
greater understanding of the way hybrid rocket fuels burn. Furthermore, previously undetectable defects, such as
cracks and voids, may be identified. The generated data can further validate proposed analytical solutions
(Cantwell, 2014) to the regression rate equations (Marxman and Gilbert, 1963) associated with hybrid rocket
propulsion systems.
Innovative Design and Construction of 3D Architecture Si/Li2S Batteries
Melissa Cronin
Mentors: Morteza Gharib, Isabelle Darolles, and Azin Fahimi
Since 1990, lithium ion batteries have ruled the portable devices industry. However, they have major drawbacks
such as their flammable electrolyte, degenerative effects, and low storage capacity. A safer battery with greater
storage capacity is needed for the upcoming market of next generation smartphones, electric vehicles, and smart
grid technology. Two materials, silicon and lithium sulfide, are promising candidates as electrodes for a new lithium
ion battery - which will have more than three times the specific energy of current lithium ion batteries. Certain
drawbacks have made the use of these electrode materials impractical, however the Gharib lab has designed a new
battery that allows the use of these electrodes. Through the use of 3D architecture, a carbon nanotube scaffold,
and a solid electrolyte, silicon and lithium sulfide's drawbacks are negated. We seek to build a prototype battery to
begin testing, and have made progress towards its completion. Both electrodes have been tested in coin cells in a
half cell configuration with the ARBIN cycling machine. The carbon nanotube lattice is grown with a Lindburg/Blue
furnace, with carbon nanotubes being grown successfully on a metal current collector over 1mm high. Experiments
are currently ongoing as to maximize the height of the grown carbon nanotubes, and to scale the carbon nanotube
up to a larger furnace. A prototype is currently in development and is expected to be done by the end of the
summer.
Testing for Paleomagnetic Wobble in Upper Cretaceous Strata
Marco Cruz-Heredia
Mentors: Joseph Kirschvink and Ross Mitchell
A new form of TPW that occurs on a short (1Myr) time frame is being investigated through the study of the
characteristic remanent magnetization (ChRM) of paleomagnetic samples from the late Cretaceous period (chrons
C33-34). To resolve the ChRM, an initial measurement of the natural remanent magnetization (NRM) is made
followed by liquid nitrogen cooling and subsequent NRM measurement. The samples are then demagnetized using
three successive decaying alternating magnetic fields with increasing strength and an NRM measurement at each
step. Finally, the samples are heated at increasing temperatures until the NRM vanishes. The NRM measurements
at the highest temperatures should form a vector that represents the ChRM and can be used to find the direction of
the ambient field, and thus the location of the magnetic north pole, at the moment of lithification. Paleomagnetic
data from the Scaglia Rossa formation in Furlo, Italy have provided very well behaved vectors that clearly show the
magnetic reversal at the start of chron C34N. This data, along with data from several locations in California, can be
further analyzed for evidence of the hypothesized 1Myr TPW phenomenon. Sufficient evidence must be found in
order to measure the statistical significance of our findings.
Orbital Period Determination of Cataclysmic Variables Using Photometric Survey Data
Daniel Cushey
Mentors: Thomas Prince and Ashish Mahabal
Cataclysmic variable stars are characterized by their short outbursts in luminosity, followed by periods of
quiescence during which many systems demonstrate a periodic luminosity. Cataclysmic variable periodicity is of
great interest due to the distribution of periods among the population which shows a distinct period minimum at 77
minutes and a gap from two to three hours in which very few systems are found. A Python pipeline was developed
to search for periodicity in cataclysmic variable systems taken from literature and five online databases using
photometric data from the Palomar Transient Factory and the Catalina Real Time Survey. The pipeline implements

16

preconditioning steps to remove extraneous data before a Lomb Scargle method is used to find candidate periods
for each system. Candidate periods were inspected using folded light curves displayed through an original HTML
interface that included multiple added functionalities to improve efficiency and accuracy. Follow up observations
were performed on selected candidate periodic systems using the Palomar 60 telescope. Further photometric and
spectroscopic observations and data analysis will be performed to confirm periods in candidate systems and further
study eclipsing or interesting sources for possible publication.
Optimizing Wireless Energy Transmitter
Bailey Da Costa
Mentors: Ali Hajimiri and Florian Bohn
Currently, designers of portable devices must strike a balance between device size and battery lifetime, or, in other
words, portability and convenience. This trade-off can be solved using wireless power transmission. By merely
being in a room with a wireless power transmitter, a laptop or phone can charge automatically. As part of previous
research efforts, Professor Hajimiris Caltech High-Speed Integrated Circuits (CHIC) group has created an array of
PCBs that are able to charge devices up to 15 ft away. However, the PCBs that make up the transmitter array have
a low frequency, overheating issues, and a significant amount of jitter, all of which decrease power transmission
efficiency. The objective of my summer undergraduate research fellowship is to test, develop, and assemble a new
transmitter board that operates at a higher frequency with less jitter. The new boards have updated amplifiers and
a flip chip design which should significantly decrease jitter and overheating.
A Homology Modeling Protocol to Predict Non-Native Intermediate States in the Cotranslational Protein
Folding Pathway
Collin Davda
Mentors: Thomas Miller and Reid Van Lehn
Cotranslational folding is the process by which some proteins fold into tertiary structure while still being translated
by the ribosome. A subset of proteins that fold cotranslationally fold into metastable, non-native conformations
while they are still partially translated. These intermediate states must be reversed or bypassed for the fully
translated sequence to fold into its native conformation, possibly due to interactions with chaperone proteins or
other ribosome-associated complexes. Previously, identifying proteins that encounter metastable intermediate
states while folding cotranslationally has proven computationally expensive, limiting these studies to peptide chains
of less than 100 residues and small sample sizes. We present a high-throughput, computationally inexpensive
method for prediction of proteins that reach metastable, non-native intermediate states in the cotranslational
folding pathway. The protocol uses homology modeling to predict structures for partial sequences truncated at
intervals of 9 amino acids from the N-terminus, analogous to protein translation. This efficient protocol allows for
rapid screening of large numbers of cytosolic proteins to identify sequences that warrant treatment with a higher
level of theory. The method may also enable the identification of general sequence or structural features shared by
this set of proteins that could provide physical insight into the cotranslational folding process.
Modification of Morpholine and Piperazine Rings in Pharmaceuticals
Ryan Dempsey
Mentors: Brian Stoltz and Alexander Sun
Linezolid is a synthetic antibiotic used for the treatment of infections from Gram-positive bacteria that are antibiotic
resistant. It is used as a last resort when other treatments like penicillin are ineffective. The morpholine ring is
present in many popular pharmaceuticals that are currently on the market. The enantioselective Tsuji reaction
developed in the Stoltz laboratory can be used to functionalize morpholine rings to modify these drugs. Using this
technique, among others, we synthesized functionalized variants of the morpholine ring. We are hoping to reduce
side effects of these drugs and/or increase their efficacy. For example, in Linezolid, we hope to eliminate relatively
common side effects of nausea, headache, and diarrhea or to make the drug metabolize more slowly so that it only
needs to be taken once a day instead of the current dosage of twice a day. We plan to determine if any of the
modified drugs chemical properties were enhanced by the functionalized through examination through molecular
docking software and by involving the biologists at the City of Hope.
Observations of Shock Structure in High Void Fraction Multiphase Flow Using X-Ray Densitometry
Nishant Desai
Mentors: Steven Ceccio and Melany Hunt
Gas-water mixtures have previously been shown to achieve a minimum speed of sound at a void fraction of 50%
and low mixture pressures. In this study, we designed a blowdown tunnel and gas injection system to produce
bubbly flow and achieve Mach numbers up to 6 in a square test section. The bubbly flow was achieved by injecting
gas into the water flow upstream of the test section. The test section static pressure and inflow velocity were
controlled by manipulating the relative pressures of the reservoir and drain tanks. A pressure reduction allowed for
low sound speed and thus made it feasible to achieve our desired Mach numbers. After construction, a deflection
wedge will be placed in the tunnel to produce an observable shock. High-speed video, pressure transducers, and X-

17

ray densitometry will be used to measure mixture pressures, bubble and liquid speeds, and actual void fraction of
the flow. These modern imaging capabilities will provide a novel look at shock structure and possibly shockboundary layer interactions in multiphase flows.
Targeting Terphenyl Diphosphine Molybdenum Compounds Bearing Metal-Ligand Multiple Bonds
Bogdan Alexandru Dimitriu
Mentors: Theodor Agapie and Joshua Buss
Molybdenum complexes featuring metal-ligand multiple bonds are utilized as model complexes for metalloenzyme
active sites and reactive metal surfaces and have been shown to act as efficient catalysts for proton reduction,
dinitrogen fixation, carbon monoxide reductive coupling, and alkyne metathesis. We present the synthesis and
characterization of a series of molybdenum compounds supported by a para-terphenyl ligand framework en route
to an unprecedented molybdenum hydride hydrosulfide complex. Hoping to access higher oxidation state
molybdenum complexes, bearing MoS multiple bonds, a related meta-substituted bis(2(diisopropylphosphino)phenyl) phenol ligand was prepared. A tri-carbonyl molybdenum complex supported by the
potassium phenolate was synthesized and its solid state structure was determined via single-crystal X-ray
diffraction. The reactivity of the newly generated compound was further investigated with the aim of establishing
molybdenum-arene interactions and ultimately forming sulfur ligated molybdenum complexes. In addition, an
anthracene-supported phosphide chloride group transfer reagent was synthesized and subsequently used in an
attempt to generate molybdenum-phosphide moieties on these ligand platforms.
The Optimization of ARPES Resolution Based on Space Charge Effect
Weiyi (Sophie) Ding
Mentor: David Hsieh
The resolution of ARPES (Angle-resolved photoemission spectroscopy) has been improved greatly over the years.
The space charge effect, which is caused the Coulomb force between photoelectrons, leads to the broadening and
shift in Fermi level. This is one of the major causes of the loss of resolution in ARPES. In my project, I am going to
explore space charge effect in terms of laser intensity and beam size. In general, low intensity and large beam size
can provide higher resolution. In this process, I am going to measure the beam size using knife-edge methods,
build a variable attenuator using half-wave plate and polarizer to tune the output power, and fit the data to a
Fermi-Dirac function to calculate the band broadening and shift.
Combining Experimental and Computational Approaches to Observing Photoactivator Chemistry in
Atmospheric Aerosols
Kayan K. Dingilian
Mentors: V. Faye McNeill and Richard C. Flagan
Secondary organic aerosols (SOA) form in the atmosphere when inorganic aerosols interact with volatile organic
compounds (VOC) in the aqueous phase. Light-absorbing organic molecules are a common substituent of
atmospheric aerosols, and there is evidence that aerosol formation may be driven by light-absorbing organics.
These molecules, known as photoactivators (PA), can lead to direct organic molecule oxidation in the formation of
aerosols. Methods of this project consisted of a combination of experimental and computational experiments as
follows. Experiments were done with a reaction chamber emulating conditions in the atmosphere with UV light.
A solution of the inorganic aerosol seed and photosensitizer were passed through an atomizer and differential
mobility analyzer (DMA) to generate and size-selected aerosol particles. This stream was added to a humidified
nitrogen flow and a stream containing a PA and run through the chamber. The reaction chamber operated at
continuous flow steady-state. The exiting particles were characterized. Quantum chemical calculations were
performed on a series of molecules proposed to be candidate PAs to observe their highest occupied molecular
orbital (HOMO) and lowest unoccupied molecular orbital (LUMO) energy levels. Results have shown the emergence
of possible PA candidates, including 4-hydroxy-3-nitrobenzoic acid, which is more effective than 4-imidazole
carboxaldehyde, an established photoactivator. These conclusions can be verified through computational
experiments with higher levels of theory.
Finding the Resonant Frequency of the Zwicky Transient Facility (ZTF) Telescope
Joe Donermeyer
Mentors: Richard Dekany and Roger Smith
The goal of this project is to ensure that the Zwicky Transient Facility (ZTF) telescopes shutters will not open with
a frequency that resonates with the telescope. Excessive vibration of the telescope will affect the image quality. I
used a linear motor to model the opening of the shutters. To test the effect on image quality, I set the telescope to
focus on specific stars and then slowly moved the telescope, taking images every 15 ms. During this time, the
motor oscillated while attached to the telescope. Using a program to find out how much the image of a specific star

18

shifted from its expected location in each time step, I determined how the profile affected the image. I ran profiles
at a range of frequencies to find which ones would affect the telescope the most. 2.5 Hz, 4.5 Hz, and 5.5 Hz were
the frequencies that stood out as having the largest effect on the image quality. This information will prevent us
from attaching shutters to the telescope that detract from its image quality.
Dissecting the Neural Circuitry of Aggression in Drosophila melanogaster
Rebecca Du
Mentors: David J. Anderson and Eric Hoopfer
Understanding the neurobiology of emotion requires knowledge of how the brain controls emotional behaviors such
as aggression, courtship and fear. Recent work with Drosophila melanogaster has shown that it is possible to
isolate the neural populations that are responsible for these complex social behaviors and map their connectivity,
thereby providing valuable insight into the neural circuitry controlling them. This SURF project seeks to use
quantitative behavioral assays to isolate small neuronal populations that are responsible for male aggressive
behaviors in Drosophila. Specifically, we will test 20 GAL4 linesdriver lines that express in specific subsets of
neurons in the brainwhich are known to promote various aggressive behaviors such as wing threat, lunging and
tussling when neurons are temperature-activated. To do so, we separate the GAL4 gene into its two functional
domains, the DNA-binding domain (DBD) and the transcription-activation domain (AD), and target them with
different promoters. We can then cross each of the 20 DBD lines to each of the 20 AD (transcription-activation
domain) lines, and only neurons in which both halves of the GAL4 gene are expressed will be able to be activated.
Preliminary tests of a small subset of the 400 potential split-GAL4 combinations have identified 4 lines which
exhibit significantly higher frequencies of lunging than the negative control, and 3 lines which show higher
frequencies of wing extension than the negative control. The next set of experiments to conduct will be to cross
these 7 lines to each of the 20 AD lines in order to further narrow down the neural population underlying these
particular aggressive behaviors. If behavioral analysis of any of these intersections produces the expected
aggression phenotype, it will also be dissected in order to image the labeled neurons with the end goal of
identifying the specific neural populations underlying various aggressive behaviors, and understanding the neural
circuitry underlying this emotional behavior.
Investigating Multi-Epoch Spectral Variability in Quasars
Alison Dugas
Mentor: Matthew J. Graham
Various astrophysical processes can only be studied in the time domain, including the accretion of matter into black
holes. Quasars vary in time at all wavelengths and are believed to be caused by supermassive black holes. The
Catalina Real-time Transient Survey (CRTS) is the largest currently-operating open transient survey, and it has
light curves for about 340,000 spectroscopically-confirmed quasars. The Sloan Digital Sky Survey (SDSS) has as
many as three spectra for each of these objects, taken across about ten years. The aim of this project is to identify
quasars with unusual spectra and highly variable light curves and investigate whether their spectra have also
changed throughout time. We have developed a way to identify statistically deviant light curves, and have used
this to find quasars that display variability both spectroscopically and in brightness.
Development of Criteria for the Analysis of Bimodal Distributions in Stochastic Biochemical
Reaction Networks
Cody Dunn
Mentors: Richard Murray, Ania Baetica, and Vipul Singhal
Gene expression is a naturally random process. Experiments in synthetic biology have shown stochastic behavior to
interfere with mRNA and protein levels in engineered biocircuits. We can use stochastic models to help engineer
biocircuits. In this research, we analyze stochastic effects in the negative autoregulation reaction network using the
stochastic modeling framework of the chemical master equation. Negative autoregulation is the process by which a
gene self-regulates; the proteins produced by that gene repress the creation of more proteins. The chemical
master equation describes how probability distributions of molecule species in a reaction network evolve as a
function of time. We are especially interested in determining the influence of reaction rates on the modality of the
steady state probability distributions. We developed criteria and implemented them in MATLAB to test whether
negative autoregulations free protein distributions were unimodal or bimodal and then attempted to find sets of
parameter values that produce bimodal stationary distributions. Our results may help forward experimental design
by offering information on the correct regions of parameter values for the desired modality of the distribution. We
plan to use the criteria for bimodality that we developed to analyze other biochemical reaction networks displaying
the property of bimodality.

19

Vernacular Music in Medieval England and Its Relationship to Modern Music

Matthew Edwards
Mentor: Jennifer Jahner
The study of medieval vernacular musicMiddle English lyrics and songsis problematic due to a frequent lack of
descriptive musical notation for the lyrics, if any notation exists at all, and also due to the scarce quantity of such
lyrics. Many facets of medieval music composition and performance remain unknown or debated, which prevents us
from knowing how accurate our recreation of medieval songs are, the circumstances under which songs were
composed or performed, and who would have composed or performed them, amongst other things. London, British
Library, MS Harley 2253 contains one of the most important collections of unnotated Middle English lyrics, but we
dont know for certain what the manuscripts purpose was. The Harley manuscript offers an essential source of
information, and various collections of musically notated Middle English lyrics provide additional information about
notation and transmission. In learning about medieval song, theres a very limited supply of historical evidence,
and so while we may never know the melodies for many songs, working backwards from more recent music and its
performance, composition, and transmission could teach us more about medieval music. So, I plan to compare
various aspects of medieval and modern music, using selected songs and lyrics.
The Effects of Bursty Star Formation Histories on Population Gradients, Stellar Kinematics, Gas
Distributions, and Dark Matter Profiles in Simulated Dwarf Galaxies
Kareem J. El-Badry
Mentor: Philip F. Hopkins
We use the FIRE suite of cosmological zoom-in simulations to study the role of stellar feedback in regulating the
evolution of dwarf galaxies. Our sample consists of 10 isolated galaxies with stellar masses between 104 and
1011 and halo masses between 109 and 1012 . The FIRE simulations use explicit models of the multi-phase ISM
and state-of-the-art treatments of star formation and stellar feedback derived directly from stellar evolution
models, with zero tunable parameters. They are run with sufficiently high force and mass resolution to resolve
galactic structures on the scale of giant molecular clouds in a cosmological setting. We demonstrate that the
dynamical and morphological properties of dwarf galaxies are highly dependent on changes in the star formation
rate on timescales of only 100 Myr. Rapid gas outflows following bursts of star formation lead to a periodically
time-varying potential, temporarily quenching star formation, driving short-timescale stellar migration, and placing
stars on highly radial orbits. The gravitational pull from outflowing gas also causes significant short-timescale
variation in the central dark matter density, which continues to oscillate after the star formation rate has stabilized.
We find that these effects are strongest in galaxies with halo masses near 1011 , likely because star formation is
inefficient at lower masses, while at higher masses, galactic potential wells are deep enough to retain cold gas
during starbursts.
A High-Order Method for Wave Propagation in 3D Dielectric Waveguides of Arbitrary Transverse Shape
John R. Emmons
Mentors: Oscar Bruno and Emmanuel Garza
Dielectric waveguides play essential roles in fiber optic devices, wireless communication, and other specialized
electronics. However, modelling the propagation of electromagnetic fields in dielectric waveguides of arbitrary
shape presents several challenges. In particular, the mode spectrum consists of a discrete and continuous set of
eigenfunctions and the continuous modes requires special treatment to avoid low order convergence. Previous
works do not fully address this challenge or ignore it altogether, limiting their use to simplified cases and making
them unsuitable for physically realistic simulations.
We present progress on a general, high-order numerical method for simulating dielectric waveguides with complex
3D geometries requiring only modest computational resources (a single workstation). The method is based on first
expanding fields into the modes of the waveguide, then propagating each mode independently. We characterize the
continuous spectrum of radiation modes and make progress toward a general prescription for its discretization.
Using this approach, we can compute a highly accurate expansion of an arbitrary field into the modes of the
waveguide, including discrete and continuous modes, yielding a more general technique for high precision dielectric
waveguide simulations.
Formal Synthesis of Switching Protocols for Estimation and Control of Aircraft Electric Power Systems
Charlie Erwall
Mentors: Richard M. Murray, Benson Christalin, and Scott C. Livingston
Aircraft electric power systems (EPS) are growing in size and complexity as more mechanical, pneumatic and
hydraulic systems are being replaced by electrical subsystems. These electrical systems are equipped with
contactors that can reroute the circuits in case of component failure. The growing complexity of aircraft EPS
increases the need for automatically generated protocols for controlling contactors, since current hand-practice
design becomes less viable for larger systems. In this project formal specifications of safety and fundamental
requirements for aircraft EPS are proposed. A Python script which automatically generates a protocol describing

20

safe, allowable control actions based on these specifications is developed and such a protocol is verified on an
aircraft EPS testbed. Implementation of these control protocols rely on the ability to estimate the state of
components, since the amount of sensors in the system is limited. This process of estimation can benefit from
contactor switching, but is restricted by the same requirements as the control. Thus, implementation of an
estimation algorithm subject to those constraints is also explored. Lastly, joint execution of estimation and control
protocols is explored and discussed.
Progress Towards the Total Synthesis of Eucomic Acid
Benzi Estipona
Mentors: Brian Mark Stoltz and Beau Pritchett
Eucomic acid is a naturally occurring compound derived from the genus Vanda. Eucomic acid has recently been
shown to serve as a global stimulus of cytochrome c oxidase in the body. This function has been directly linked
with attenuating the cell aging process as well as serving as a preventative measure towards cancer. Herein, we
utilize synthetic methodologies pioneered by the Stoltz group, namely, the asymmetric allylic alkylation of
dioxanone substrates, to achieve a total synthesis from commercially available cyclohexanone dimethyl ketal to
access enantioenriched alpha-hydroxy methyl esters, which serves as the basic framework for many biologically
active compounds including eucomic acid. Currently, we have been successful in synthesizing the dimethyl ester
intermediate leaving only saponification of the diester to yield the natural product.
Development of an SFC-ESI-MS System for Online Observation of Prebiotic Chemical Reaction Products
in Supercritical Carbon Dioxide via Mass Spectrometry
Jieyuan (Jenny) Fan
Mentors: Jesse L. Beauchamp, Daniel A. Thomas, and Kathleen T. Upton
Beyond well-known applications to green chemistry, supercritical carbon dioxide (scCO2) provides a unique
opportunity to investigate chemical reactivity in an environment that differs from water and organic solvents.
Hypothesized to be present in the subsurface of several planets, both within and beyond our solar system, scCO2 is
therefore an intriguing medium in which to explore prebiotic chemistry and the possibility that it may have
supported nonaqueous life. The motivation in studying prebiotic chemistry in scCO2 is two-fold: in addition to the
possibility that it may have contributed to the emergence of life on earth, it will help to inform the search for life on
other planets. To examine reactions in scCO2, a high-pressure reactor coupled to a mass spectrometer has been
developed. The system comprises a high pressure pump to generate scCO2, a reaction vessel enclosed in a
temperature-controlled oven, and a modified ESI source that interfaces the high-pressure system with a Thermo
Scientific LTQ-XL ion trap mass spectrometer for real time online monitoring of reactants and products. Diagnostic
testing with caffeine and other molecules known to be soluble in scCO2 has been performed, and studies of
chemical reactivity, including the synthesis of amino acids and peptides from simple precursors, are underway.
Spectral Action for Bianchi-Type IX Minisuperspace
Wentao Fan
Mentor: Farzad Fathizadeh
We carry out a detailed study of the Seeley-de Witt coefficients appearing in the asymptotic expansion of the
spectral action for the Dirac operator acting on sections of the spin-bundle over the Bianchi-type IX
minisuperspace. The first three coefficients are calculated by expanding the heat kernel and applying parametric
pseudodifferential calculus. We have also developed an efficient general method for computing the Seeley de-Witt
coefficients associated with an elliptic positive operator on a compact manifold, which is based on making use of
Wodzicki's noncommutative residue. The results derived from both methods are compared to confirm the validity of
the new approach. In addition to being remarkably efficient, after taking advantage of the SU(2) symmetry of the
Bianchi-type IX metric, this method yields an elegant proof of a rationality result. That is, we show that a general
term in the spectral action for Bianchi-type IX metric is expressed by a several variable polynomial with rational
coefficients evaluated on the cosmic evolution factors 1 (), 2 (), 3 () and their higher time derivatives of a
certain order. An especially interesting family of metrics called Bianchi IX gravitational instantons, which, after a
correct choice of a conformal factor, form a family of self-dual Einstein metrics, are then considered and the
rationality result is extended to this case. Considering the latter and the fact that there is a parametrization of
Bianchi IX gravitational instatons in terms of theta functions with characteristics, arithmetic and modular properties
of their spectral action and their relation with modular forms are studied in detail.
Optimal Trajectory Generation Under Environmental Uncertainties Using Signal Temporal
Logic Specifications
Nuno Ferreira Duarte
Mentors: Richard Murray and Samira Farahani
Signal Temporal Logic (STL) has successfully been used for specifying a wide range of behaviors for systems and
environments. Previous works using Linear Temporal Logic (LTL) have synthesized controllers when specifying
properties of discrete time signals but are incapable of handling continuous signals with time dependency, in

21

particular, when optimizing trajectories under environmental uncertainties, such as static or dynamical obstacles,
known and unknown obstacles. To address these issues, we use STL and our aim is to design a reactive strategy
and trajectory planning for a surveillance mission in a partially known environment; the autonomous vehicle will
satisfy tasks such as periodically monitoring areas, path planning, staying safe, and avoiding obstacles while the
position of some of them is not known a priori. We also compare the performance of the obtained controller with
the one obtained using LTL language.
Developing a Search for Dark Matter Direct Production Using Razor Variables in Proton-Proton
Collisions at 13 TeV
Jared Filseth
Mentors: Harvey Newman, Maria Spiropulu, Cristian Pea, and Javier Duarte
Dark matter makes up 27% of the mass-energy budget of the universe, but still remains elusive today. In recent
years, searches for dark matter production in proton-proton (pp) collisions have been carried out up to a center of
momentum energy of 8 TeV. This project involves improving and reengineering the analysis methods used on 8
TeV data, so that they can be applied to new 13 TeV data. These methods involve looking at razor variables, which
are used to quantify the imbalance in transverse momenta in each collision. In order to update these methods, I
have been measuring quantities such as the trigger and lepton efficiencies for simulated events as well as Compact
Muon Solenoid (CMS) data. Further studies would involve finishing the search, using these methods, once a
sufficient integrated luminosity has been recorded.
Written Roles: Female Letter Writing in Romantic England
Jennifer Frazin
Mentor: Kevin Gilmartin
In 18th and 19th century England, written correspondence provided the average female with a means to extend
herself beyond the domestic sphere of her own home, but did this extension entail greater independence? In order
to answer this question, this project turns to 18th century letter writing manuals, episodes of correspondence in
three Jane Austen novels, and Mary Shelleys Frankenstein, as well as a collection of published letters by Mary
Wollstonecraft. Analyses of these works and related scholarship serve to place the Romantic woman in relation to
her world and, in so doing, pose several other key questions. What was the Romantic womans role, and how did
letter writing reflect, change, and/or solidify this position? What can we learn about expectations for female
behavior from letter writing manuals or episodes of correspondence in novels? How might these examples compare
to published collections of letters? As might be expected, answers to these questions are anything but black-andwhite. My analysis demonstrates how the female letter writer can be viewed both as oppressed, by her limited
means of representation, and liberated by the opportunity to transcend private domestic life. Analytical recognition
of this ambiguity paves the way for future literary and historical analysis of this topic.
Investigation of the Role of Treslin in DNA Replication in Xenopus Egg Extracts
ngel Glvez Merchn
Mentor: William G. Dunphy
Once every cell cycle, genetic material duplicates in an accurate process called DNA Replication. In eukaryotic
organisms, replication involves the ordered recruitment of conserved proteins on replication origins. This assembly
culminates with the activation of the replicative helicase, which unwinds the double strand and gives rise to the
onset of DNA replication. However, despite its great importance, our knowledge on this process in vertebrates is
limited. In this project, we aimed to improve the current understanding of DNA replication onset using Xenopus
laevis egg extracts as a model system. Performing immunodepletions, we removed essential origin proteins such as
Treslin, RecQ4 and TopBP1. Subsequent chromatin-binding assays allowed us to develop spatial-temporal models
on proteins behavior at replication origins. Additionally, time course origin-binding experiments of crucial players
were performed in untreated and drug-treated conditions, giving us insights into the complex dynamics of the
process. These results led us to contemplate a potential extra role of one of the proteins in the intra-S phase
checkpoint, although additional experiments need to be carried out to confirm this. In conclusion, this project has
shed some light on the poorly understood DNA replication onset, and might have unveiled a new role of one of its
players as a safeguard of genomic stability.
Multiplexed Cell Sorting Based Upon Relative Quantitation of Endogenous mRNA Targets by in situ
Hybridization Chain Reaction
Siva Gangavarapu
Mentors: Niles Pierce and Aneesh Acharya
Hybridization chain reaction (HCR) has been recently developed as an alternative method of signal amplification for
in situ hybridization systems. It can also be applied to cell sorting using genetic markers present in various cell
types. This project aims to demonstrate HCR as an effective system for studying endogenous mRNA targets and
isolating and sorting cells. We picked targets picked from four different tissue types: retinal g protein coupled

22

receptor from the retina, surfactant protein A1 from the lung, albumin from the liver, and chymotripsinogen B1
from the pancreas. Results have demonstrated both very high signal from targeting the surfactant protein in lung
tissue and low background signal. By performing redundancy studies using multiple probe sets, probe signal in all
samples could be increased by eliminating poorly performing probes. Moreover, we will be able to demonstrate
relative quantitation between HCR signal and mRNA concentration. After demonstrating high presence of a target in
the intended tissue type and low presence in the other tissue types by observing signal-to-background ratios, we
can conclude that HCR can be used to isolate and sort cells from heterogeneous tissue samples using mRNA
targets.
Determining Survey Completeness for Friends of Hot Jupiters
Paige Gannon
Mentors: Heather Knutson and Henry Ngo
Some systems containing hot JupitersJupiter mass planets orbiting extremely close to the host staralso host
stellar mass companions, which may have influenced the migration of the planets. The Knutson groups standard
method for examining these systems is to use Kecks NIRC2 camera with Kecks adaptive optics (AO) to directly
image their targets. There is an alternate survey mode, angular differential imaging (ADI), typically used with a
coronagraph to occult most of the primary stars light. It is capable of resolving fainter companions, though it takes
more observational time and places a lower limit on projected separations. Through injecting synthetic companions
into the target systems, I determine the expected difference in the number of companions found by the current
survey mode and the ADI survey mode, and compare it with the percentage of companions missed due to their
geometrical arrangement with the target stars. Through this analysis, I have also created an automatic companion
detection algorithm, which should provide a useful check for manual inspection.
Analysis of Chromatin Conformation of Cis-Regulatory Modules in Drosophila melanogaster
Galen Gao
Mentor: Angelike Stathopoulos
Spatial and temporal control over gene expression is often mediated by sequential action of cis-regulatory modules
(CRMs)stretches of DNA usually several hundred base pairs in length near the genes they regulate. Recently, it
has been hypothesized that local changes in chromatin conformation can reposition CRMs and that these shifts may
cause changes in gene expression. One such gene, brinker (brk), encodes a transcription factor essential for proper
Drosophila dorsal-ventral development and is controlled temporally and spatially in early development by a 5 CRM,
a 3 CRM, and a promoter proximal element (PPE) near the brk coding region. To test the role of the relative
positioning of these elements in brk expression, fluorescent probes marking the 5 CRM, 3 CRM, and PPE were
introduced into nuclear cell cycle 14 embryos, which were then imaged via fluorescence microscopy. A spot
detection algorithm employing a multi-scale Laplacian of Gaussian detector with automatic scale selection was used
to detect the probes positions and group them into triplets corresponding to each nucleuss conformation of 5
CRM, 3 CRM, and PPE components. The resulting triplets were classified into eight conformations based on their
elements proximity to each other and then mapped back onto the embryo.
Preparation of Isoquinolines for Rapid, Convergent, and Enantioselective Access to Jorumycin
Kevin Gao
Mentors: Brian Stoltz and Max Klatte
Accessing diverse isoquinolines has important implications for pharmaceutical applications and natural product
synthesis. These structural motifs are found in numerous potential antitumor agents among the
tetrahydroisoquinoline (THIQ) alkaloid family. The Stoltz group has a longstanding interest in the total synthesis of
biologically active THIQs. Recent efforts towards jorumycin have been a particular research focus. Our original
synthetic strategies were built on aryne annulation. Issues with scalability, efficient material throughput, and the
demethylation of key substituents in the late stages of the synthesis led us to seek alternative access to
isoquinolines. In this effort, we employed a robust approach utilizing known cross-coupling methodology such as
the Heck and Sonogashira reactions to generate our desired substituted isoquinolines from readily available
3,4,5-trimethyoxybenzaldehyde. Benzyl substituents were installed to enable facile deprotection. Fagnou coupling
will allow access to the carbon framework of jorumycin.
Characterization of CZT and CdTe Pixel Detectors for Future Astrophysical X-Ray Missions
Daniel Gawerc
Mentors: Fiona Harrison, Hiro Miyasaka, and Vikram Rana
The Nuclear Spectroscopic Telescope Array, or NuSTAR, has CdZnTe (CZT) hybrid detectors in its focal plane for
x-ray observation. Next generation detectors can benefit from NuSTAR analysis. High quality CZT crystal yield is
low. Studying CdTe crystals is important, as they were found to have a higher yield than CZT. CdTe crystals are
crucial for wider focal plane missions, where more crystals are necessary. CZT and CdTe properties have different
upsides and downsides for x-ray detection. Leakage current shot noise fluctuations may be incorrectly read out as
x-ray signals in the detected spectrum. It is due to an applied bias voltage and the internal resistance of the crystal

23

pushing a small current into the ASIC. It increases with HV, temperature, and crystal impurities. CZT was
confirmed to have a constant leakage current over time. CdTe electrode contacts were compared to ascertain
which exhibited the lowest leakage current. Gamma flood, or radioactive source illumination, tests are used to
characterize the structure and uniformity of a crystal. 241Am was used to characterize CZT hybrid H79 by comparing
the emitted and detected spectra. Additionally, the NuSTAR laser metrology optics tracking system power decay
was estimated as a function of time and intensity.
Calorimeter Photodetector Studies for the Mu2e Project
Hallmann skar Gestsson
Mentors: David Hitlin and Frank Porter
The neutrinoless decay of a muon into a monoenergetic electron is an example of a charged lepton flavor violation.
The Mu2e project will study this process by trying to detect this decay with a proposed sensitivity of 6.7 1017 . This
will provide data that will show which types of new physics models are favorable. Produced muons will be captured
and guided by a series of superconducting solenoids onto an aluminum stopping target. Conversion electrons will
emerge from the target and enter a tracker and then a calorimeter. The calorimeter provides secondary and
independent measurements of converted electron energies. It will make use of scintillating crystals and
photodetectors to do these measurements. The crystal that is investigated in this paper is BaF2 which has a fast
and slow time component in the UV spectrum. The photodetector to be used must be able to work in a 1 Tesla
magnetic field, have high quantum efficiency in the UV spectrum, and be able to block out the slow time
component. Measurements carried out on large area avalanche photodiodes that are delta doped are presented in
the paper and determined if they meet the demands of the project.
Explicit Orbit Construction for the Matsuki Correspondence for the Affine Grassmannian
Iulia Gheorghita
Mentor: Xinwen Zhu
For a connected reductive complex algebraic group G with Cartan involution , real form G_R with respect to
conjugation , maximal compact subgroup G_c, let be the composition of and . Let K be the fixed points of
and K_c be the maximal compact subgroup of both G_R and K. If B is the flag variety of G, G_R and K act on B
with finitely many orbits. The Matsuki correspondence states that for each G_R-orbit O_R in B, there exists a
unique K-orbit O_K such that their intersection is a single K_c orbit and thus, this correspondence gives an orderreversing isomorphism between G_R\B and K\B. In 2004, D. Nadler extended this correspondence to the case
where B is replaced with the affine Grassmannian, Gr = G(C((t)))/G(C[[t]]), the polynomial loop group LG_R is
taken instead of G_R, and K(C((t))) instead of K. We seek to explicitly describe the orbits of which D. Nadler gives
a parameterization, by making use of the natural stratification of Gr.
Resonant Production of Sterile Neutrinos in the Early Universe
Lauren Gilbert
Mentors: George Fuller and Christian Ott
This study examines the cosmological impacts of a light resonantly produced sterile neutrino in the early universe.
Such a neutrino could be produced through lepton number-driven Mikheyev-Smirnov-Wolfenstein (MSW)
conversion of active neutrinos prior to big bang nucleosynthesis (BBN). During BBN, the neutron-proton ratio
depends sensitively on the electron neutrino flux. If electron neutrinos are being converted to sterile neutrinos, this
makes the n-p ratio a sensitive probe of possible sterile neutrinos. We therefore use experimental measurements
of primordial Yp and D/H to place limits on this process.
The Effect of Chronic Nicotine Use on Neurons in the Substantia Nigra in Relation to Parkinson's
Disease
Heather Gold
Mentors: Henry A. Lester and Beverley Henley
The chronic use of nicotine has been suggested to have a neuroprotective effect against Parkinsons disease by the
inverse association between cigarette use and Parkinsons disease incidence. This project seeks to understand the
mechanisms behind this effect and specifically study the effect of nicotine on neurons in the substantia nigra, the
brain region in which neurons selectively die in Parkinsons disease. After chronic exposure to nicotine, the
transcriptome of a neuron was sequenced using single-cell RNA-sequencing, allowing data libraries to be collected
on individual neurons from the heterogeneous population of the substantia nigra. After sequencing, each library
was aligned against a genome, quantified, and normalized before differential analyses were performed. Data
processing was performed with Tophat, Bowtie, htseq-count, and Cufflinks, as well as custom python scripts.
Preliminary results indicate the suppression of neuroinflammation and cytokine signaling as possible
neuroprotective mechanisms, but further testing must be conducted to verify and replicate these results with
expanded datasets.

24

Ground Deformation in Simulated Earthquake Shaking Using Complex Fourier-Based Amplification

Factors
Carlos Gonzalez
Mentors: Domniki Asimaki and Jian Shi
It has long been known that local site characteristics have notable effects on the ground motions caused by an
earthquake. Most modern U.S. seismic design codes for building structures account for the specific near surface
response of the soil layers in the location of interest. However, the established site-classification criteria in use
today do not sufficiently describe soils susceptibility to nonlinear ground motion. Previous work has been done
validating a proposed nonlinear model's (NLHH) ability to construct the Fourier amplitude spectrum for various
sites. This project seeks to collect data and perform statistical analysis on the viability and accuracy NLHH's ability
to construct the Fourier phase curve. Using data from the KIK-net downhole array, we are able to compare the
performance of various simulations ability to recreate a ground motion time series Fourier phase curve.
Additionally, we have explored an empirical relationship between the transfer functions for Fourier amplitude and
Fourier phase for linear, homogeneous, viscoelastic soil layers. Once verified, we can explore it further for more
complicated soil layers with the goal of using it for post-facto broadband ground simulations to verify simulations
ability to predict ground motion based on nonlinear site characteristics.
Mechanistic Studies on QUINAP and Its Triflate Precursor
Ashay Makarand Gore
Mentor: Scott Virgil
A novel method of the asymmetric synthesis of the chiral ligand QUINAP via dynamic kinetic resolution was
recently developed. This synthesis involves an isomerization process of the arylpalladium intermediate in which the
isoquinoline piece can adopt two positions in relation to the naphthalene ring. Mechanistic studies to deduce the
mode of isomerization of the arylpalladium intermediate and the triflate precursors via deuterium labeling and
kinetic analysis were performed. Work on isolating the arylpalladium intermediate was also conducted.
A Novel Radiation Shield Design for sub-K, Angle-Resolved Photoemission Spectroscopy
Arjun Goswami
Mentors: David Hsieh and Tejas Deshpande
Angle-resolved photoemission spectroscopy (ARPES) is a technique that images the electronic bandstructures of
solids by inducing photoemission in a sample, and analyzing the energy and momentum of the ejected
photoelectrons. The Hsieh lab's laser-based ARPES system at Caltech cools down samples to sub-K temperatures
under ultra-high vacuum (UHV) conditions, enabling measurements of condensed matter phenomena. Such
measurements may provide insight into the existence of topological superconductors, a potential platform for lowdecoherence topological quantum computation. To achieve proper measurement conditions, liquid cryogens
(nitrogen and helium) are used to cool down the sample, which must be encased in radiation shields during this
process. Here, a novel design for a radiation shield in the form of a machined helical spring is presented.
Characteristics that are embodied in the design include UHV-compatibility, a robust thermal enclosure for the
sample, and a mechanism for sample loading. Analytical techniques, as well as numerical techniques (finite
element analysis) were used to quantify mechanical and thermal properties of the shield. After testing the radiation
shield on the system, modifications and refinements can be recommended for optimal functionality.
Investigation of Human Magnetoreception Using Bistable Visual Illusions
Mara Green
Mentor: Joseph Kirschvink
It is already well documented that a wide variety of organisms sense and respond to the Earths magnetic field.
Examples of bacteria, invertebrates, birds, fish, reptiles, and even mammals use this magnetoreception for
navigation and orientation. Since humans likely descend from a magnetoreceptive ancestor, we are investigating a
possible latent magnetic sense in humans. Our experimental setup consists of a large Faraday cage in which we
can precisely control the magnetic field, allowing us to expose subjects to changing fields and to try to elicit
subconscious subject responses. One approach is to rotate the field in the horizontal plane and simultaneously
instruct the subject to view a bistable visual illusion. This type of illusion provides sensory information which is
equally likely to be interpreted in either of two ways, and which causes random perceptual switching between these
two interpretations. In this case, the illusion may be interpreted as rotating either left or right. We hope to
modulate subjects perception of this illusion using the rotation of the ambient magnetic field.

25

Machine-Readable Protocols and Rapid-Prototyping for Synthetic Biology Research: Implementation of

a Liquid Handling Robot
Carl Marcus Greiff
Mentors: Richard Murray and Scott Carlton Livingston
This project is focused on creating a liquid handling robot for use in the synthetic biology lab. The goal is to
improve the consistency of the TX-TL cell free expression system by reducing the potential for human error during
production, and also free up time for the researchers, increasing the total research volume of the Keck synthetic
biology lab. This is done by rapidly prototyping the robot while maintaining a high level of reusability and
versatility. For these purposes, the project is divided into four subsections. The first is to create a Python based
graphical user interface (GUI); the second is to make use of the robot operating system ROS to enable the
interaction between the GUI and the controller; the third is to implement the controller and set up the sensors and
motors of the robot; and the fourth is to help with the hardware design in order to ensure compatibility between
the hardware and software.
Predicting the Structures of Family 2 Human Olfactory Receptors
Daniel Gu
Mentors: William Goddard, III, and Soo-Kyung Kim
The sense of smell is mediated in humans by the human olfactory receptors (hORs), a group of G-protein coupled
receptors (GPCRs). Since these proteins do not crystallize easily, there are very few structures known for GPCRs.
The overall aim of the project is to use the GEnsemble methods previously developed by the lab to predict
structures and the ligand docking behaviors of these hORs. In particular, my project focused on the family 2 hORs
OR11H4, OR11H7, and OR10J5 and their docking behaviors.
The Invariant Theory of the Order 3 Automorphism of so(8)
Jeffrey Gu
Mentor: Xinwen Zhu
The Lie algebra so(8) is unique since it is the only Lie algebra to have an order 3 automorphism. This
automorphism induces a grading so(8) = g0+g1+g2, where gi is the eigenspace corresponding to i, where is a
primitive 3rd root of unity, and the set of fixed points g0 is the g2 Lie algebra. We seek to describe the action of g0
on g1 and its invariants, such as the nilpotent cone and nilpotent orbits.
Analysis of DNA Charge Transport in E. coli DNA Repair Proteins Using CRISPR Interference
Stephanie Gu
Mentors: Jacqueline K. Barton and Andy Zhou
Various environmental factors induce mutations within an organisms DNA, which can result in cancer if left
unchecked. An organism needs to employ DNA repair mechanisms to protect the integrity of the genome for itself
and future generations. Here we constructed a CRISPR interference (CRISPRi) system to study how DNA-mediated
charge transfer (CT) facilitates signaling amongst E. colis three DNA repair proteins: EndoIII, MutY, and UvrC.
Using Gibson assembly, we constructed a CRISPRi one-plasmid system that consists of two parts: an inducible
dCas9 endonuclease regulated by pBAD and a constitutively expressed single-guide RNA (sgRNA) sequence using a
GAPDH promoter. Following its construction, we will conduct an in-vivo UV sensitivity assay to examine how
expression levels of the three DNA repair proteins affect the cells ability to repair UV-induced DNA damage. We will
also use atomic force microscopy (AFM) assays to measure protein redistribution on DNA with or without base pair
mismatches. Cyclic voltammetry will be used to examine the repair proteins redox potential. These results will
verify the CT-mediated mechanism for identifying DNA damage and will establish DNA CT as a method for signaling
within biological chemistry.
Matching Tumor Antigens to T Cells
Katherine Guo
Mentors: James R. Heath and Songming Peng
Cancer immunotherapy involves stimulating the bodys immune system to fight cancer. One application of
immunotherapy is adoptive cell transfer (ACT), which involves isolating T cells from patients and picking the ones
that react to cancer. They then multiply the cells, stimulate them with immunotherapy drugs that ramp up T cell
abilities, and inject them back in the patient. While promising, a big challenge is to broaden its reach by identifying
new tumor antigens that will guide the T cells to specifically kill the cancer cells while sparing the healthy cells.
Current immunotherapy methods mainly target the common antigens, however, by discovering neo-antigens
specific to a persons cancer, the ACT treatment will be more effective. With a library of potential antigens derived
from tumor sequencing data, we aim to pull down single T cells with PDMS microfluidic chips, read the barcode of
their associated tumor antigen through sequentially coded fluorescent dyes, and then isolate T cells that can
recognize a patients known tumor antigens. By identifying a tumor antigen and its matched T cell receptor at the
single cell level, we can pave the way for personalized cancer immunotherapy.

26

Using Lens Cuttings to Derive Upper Bounds for Incidence Problems

Songshan (Bella) Guo
Mentor: Adam Sheffer
Lens cutting is a technique used to improve by Aronov and Sharir to improve the current bound for point circle
incidences involving circles of arbitrary radii in the plane. This paper used archaic partitioning techniques, so we
improve their bound by using the polynomial partitioning method, which was recently developed by Guth and Katz.
Other problems examined included the cutting number, or the cuts needed in order to convert an arrangement of
curves to an arrangement of pseudosegments, such that there are no K2,2s, of various objects, such as ellipses,
graphs of polynomials, and general polynomials.
Immobilization and Study of Fluorinated Molecular Catalysts on Graphitic Surfaces
Ayush Gupta
Mentors: Harry B. Gray and James D. Blakemore
Interfacing well-defined molecular catalysts with electrode surfaces is a key step toward constructing devices for
selective solar-fuel production. Noncovalent interactions between fluorinated aromatic cycles and other aromatic
cycles have been observed in a variety of compounds. Most notably of these is the interaction between
hexafluorobenzene and benzene. We now report synthesis of a new bipyridine ligand appended with two
perfluorobiphenyl groups. The ligand was synthesized by reacting 4,4-Dimethyl-2,2-bipyridine with lithium
diisopropylamide to afford the reduced bipyridine. This was then reacted with decafluorobiphenyl to form the final
compound. The ligand was then reacted with a rhodium cyclopentadiene dimer to form the proton reduction
catalyst that was studied. To better study the surface-attached catalyst. A high surface-area carbon material was
prepared on highly oreiented pyrolytic graphite using Ketjen black and a conductive polymer. Preliminary studies
with the carbon material show a wide range of stable potentials that can be applied. Further data about the
attachment and stability of the catalyst will be provided once the experiments are completed.
Estimation of Hypocenter for a Given Seismic Event
Gaurav Gupta
Mentors: K. Mani Chandy and Julian Bunn
The earthquake monitoring system consists of large widespread sensor network which measures the acceleration at
the ground due to shaking and sends this data to a server, and various statistical algorithms are used to detect and
issue warnings about the impending seismic activity. This paper contains the algorithm for detection of Hypocenter
using the large amount of data streams collected by Community Seismic network (CSN) sensors during an
earthquake event. The estimation of hypocenter will enable us to determine the arrival time of seismic wave using
statistical models and will contribute towards development of early warnings systems for possible damage control
during such an event.
High Sensitivity CMOS Potentiostat for in vivo Distributed Wireless Sensing
Albert Gural
Mentors: Azita Emami and Manuel Monge
Addressable Transmitters Operated as Magnetic Spins (ATOMS) are micro-scale devices capable of power
harvesting and communication at magnetic field-dependent frequencies. Their locations can be discerned by
applying a magnetic field gradient and communicating with them at correspondingly shifted frequencies. A
potentiostat, an electrochemical sensor that measures the concentration of any given electroactive chemical
analyte in solution, can be added to these devices to allow for minimally-invasive wireless implantable chips that
are capable of determining the chemical constituents of the surrounding fluids. For example, by placing many
potentiostat-functionalized ATOMS chips throughout the brain, global real-time dopamine recording can be
achieved. We present a novel CMOS potentiostat that adheres to the power, voltage, and size constraints of ATOMS
technology while still allowing for high dynamic range sensitivity and a wide redox voltage swing. The potentiostat
was designed using predictive technology models of 45nm CMOS technology and operates at 1V.
On the Bounds of the Carbery Rectangle Problem
Daniel Guth
Mentor: Nets Katz
In attempting to extend a theorem on sublevel sets to higher dimensions, Carbery, Christ, and Wright observed
that any measurable set E in the unit square in R2 not containing the corners of an axis-parallel rectangle with area
1

greater than has measure bounded by O( log( ) ). Under certain conditions the logarithm in this bound can be

removed, but it is an open problem whether or not this bound is sharp in the general case. We examine the
geometric properties of a specific case where part of the set is E is known.

27

Investigating the Role of Cellulose in Morphogenesis of Arabidopsis Hypocotyls and Shoot

Apical Meristem
Melina Theoni Gyparaki
Mentors: Elliot Meyerowitz and Arun Sampathkumar
Cellulose is synthesized at the plasma membrane by rosette-like cellulose synthase complexes (CESA). Primary cell
wall formation requires CESA1, 3 and either CESA2, 5, 6 or 9, while secondary wall formation requires CESA4, 7
and 8. The deposition and guidance of the CESA complexes are mediated by the microtubule cytoskeleton. The
Meyerowitz lab has shown that mechanical stress acts as an instructing signal to regulate cell wall synthesis at
subcellular and tissue scales via the microtubule cytoskeleton network. However, the molecular aspects of CESA
gene regulation have not been fully described with regards to mechanical forces in different tissues and cell types.
Cytokinin is a hormone involved in shoot apical meristem formation among other processes. The aim of this project
was to investigate the interactions between the CESA genes, the microtubule network and cytokinin and how they
are all related to cellulose synthesis as well as hypocotyl and meristem development. A variety of techniques was
used including In-Fusion cloning of CESA1 and CESA3 genes, next-generation in situ hybridization, imaging of
cytokinin markers in hypocotyls as well as mechanical perturbation coupled with confocal microscopy. Our results
demonstrate a putative interaction between cytokinin signalling, microtubules and cellulose synthase gene
regulation.
Preventing Environmental Release of Genetically-Modified Gut Bacteria With a Thermally-Activated
Kill Switch
Gloria Ha
Mentors: Mikhail Shapiro and Mohamad Abedi
The human body carries about 100 trillion microorganisms in its intestines, a number ten times greater than the
total number of human cells in the body. These organisms play a major role in maintaining our health and mood,
and are targeted by engineered bacteria and medicines. Unfortunately, therapeutic bacteria engineered to modify
the gut microbiota are eventually excreted into the environment, where they continue to propagate. In order to
design engineered gut bacteria that are incapable of spreading into the environment, we used TlpA, a coiled-coil
protein found in Salmonella typhimurium, to establish heat-modulated bacteria that can survive at gut
temperatures (37C) but not outside (25C). To this end, we targeted the production of D-alanine, an amino acid
that is essential for cell membrane formation, and thus for bacterial growth. Alr-100, an essential gene for Dalanine production, was made dependent on TlpA binding, and auxotrophic bacteria incapable of D-alanine
production were transformed with these plasmids. Future work will use the optimized constructs to design bacteria
that are activated at fever temperatures (40C) but not at body temperature (37C).
Reproducibility in Cancer Biomarker Studies
James Ha
Mentors: Ashish Mahabal and Thomas Fuchs
The ease of use and availability of next-generation sequencing and microarray platforms has resulted in a deluge of
cancer biomarker studies being published. Given the importance of the various clinical uses of cancer biomarkers,
including early detection screening, the development of effective treatment regimens, and estimating patients
prognoses, it is important to ensure that the results of biomarker studies are reproducible. Of the many levels of
reproducibility that may be examined in a biological study, we elected to examine the reproducibility of figures
from processed data. We addressed this reproducibility issue by building a web application that allows users to
visualize their data using a variety of methods (such as principle components analysis, correlation heatmaps, and
pairs plots), and download plots generated by the app.
Characterizing Entangled Strategies for Nonlocal Prover Games
Nick Haliday
Mentor: Thomas Vidick
The goal of this project is to broaden understanding of entangled strategies (allowing shared entangled states but
no communication) for nonlocal prover-verifier games beyond the well-understood case of XOR games, either by
generalizing known results or finding pathological examples. We constrain ourselves to 2-prover games with
classical verifiers, and focus on the values c q sdp ns (the classical, entangled, SDP, and non-signaling
values), i.e., the maximal winning probability for strategies of various sorts. We consider the last two mainly
because they are computationally and analytically tractable, and provide a route toward analyzing the first two. We
begin with marginal XOR games, an extension of XOR games where certain answers are more heavily weighted for
particular players. We show that the identity game, played on a graph, always satisfies ns = c, so that
entangled strategies offer no benefit over classical. This generalizes a result from the literature where the graph
was the path graph. We consider k-coloring games for small k, and attempt to sharpen known bounds on the c-q
gap for general games for this particular class.

28

Implementing Coarse-Grained Atomistics for Multi-Scale Material Modeling

Andrew Han
Mentors: Dennis Kochmann and Ishan Tembhekar
It is of great interest in materials science to acquire a fundamental understanding of the physics of materials across
various scales, so that a bottom-up approach to engineering material properties may be possible. Most
computational models which describe material phenomena are, while accurate, constrained to discrete length and
time scales, and these models are not capable of predicting the cascading effects of phenomena across scales.
Magnesium, a material of high interest for transportation and defense applications due to its high strength to
weight ratio, is one example of a material which requires multi-scale modeling for accurate property prediction.
Magnesium is inherently susceptible to the nanoscale phenomena of twinning, and modeling this behavior
harmoniously with atomic scale dislocation activity is computationally expensive and difficult. A new quasicontinuum method uses coarse-grained atomistics to approach multi-scale modeling by allowing simulations with
atomic resolution and finite element efficiency. This model can describe both twinning and dislocation activity in
stressed Magnesium, demonstrating the models potential to facilitate significant progress in material property
prediction and engineering.
Characterization of a Putative [4Fe-4S] Complex in UvrC by Means of X-Ray Crystallography
Sirus Han
Mentors: Jacqueline Barton and Rebekah Silva
DNA-mediated charge transport (DNA CT) is the fidelity-dependent ability of DNA to transfer electrons through the
aromatic systems that run up the helices. It has been demonstrated via a combination of genetic and molecular
assays that two proteins, Endonuclease III and MutY, participate in base excision repair, facilitated by DNA CT.
Both of these proteins contain an uncommon [4Fe-4S] cluster which until recently was thought to be only
structural in function. However, since these clusters are only redox active at physiological potentials when bound to
DNA, it is hypothesized that they aid the repair proteins in locating damage.
The Barton Group has identified UvrC as another protein that participates in NER that may contain a [4Fe-4S]
cluster. Strong evidence for the clusters existence in UvrC was obtained through EPR. In order to confirm the
clusters existence, UvrC is in the process of being purified and crystallized. Once obtained, these crystals of UvrC
will be subjected to X-Ray Crystallography, by which a crystal structure will be obtained.
An Investigation of Two Different Power Distribution Options for Small Fuel Cell Vehicles
Jacob Harmon
Mentor: Guillame Blanquart
The purpose of this investigation is to evaluate the pros and cons of two popular power distribution systems in
small fuel cell vehicles. The first system, used primarily in mid-sized vehicles, powers the drive train, while sending
excess power to battery and superconductor, theoretically increasing fuel cell efficiency. The second system
bypasses the extra battery banks, thus being a simpler, lighter, and cheaper system that is only viable in small
vehicles. Modifications to a small electric vehicle constructed by Caltech Sustainable Vehicle Club, along with
manufacturer specifications of commonly used fuel cells, provide data to be used in the analysis of the two
systems.
Using Machine Learning to Develop Algorithmic Trading Strategies for Stock Market Investing
Christopher Hazard
Mentor: Ben Gillen
Wall Street traders seek to extract information from historical market data and utilize it to identify and exploit
patterns to execute profitable trades, leading to the development of a variety of forecasting and trading strategies
ranging from simple methodologies to complex trading systems. Many approaches for achieving winning trading
strategies have their roots in information theory, mathematical finance, stochastic modelling, and machine
learning. The underlying question of this study is: Can one achieve consistent above average returns based on
recent stock price movements alone, or is other information necessary to make consistent algorithmic trading
strategies? This project attempts to utilize machine learning techniques in conjunction with price derived indicators
taken from technical analysis to predict the future direction of stock price changes by applying them to minute by
minute historical stock data. The performance of these algorithms is measured by comparing them to a benchmark
buy and hold strategy for the same securities over the same time period. Further study will include refining the
more promising machine learning derived indicators and utilizing them to create better algorithmic trading
strategies.

29

Classifiying RR Lyrae Variable Stars Using Random Forest

Kevin He
Mentors: Judith G. Cohen
RR Lyrae stars are blue, variable stars that can be found in the outer halo of the Milky Way. Because of their
variability and blueness, which arises from them being relatively hot stars, RR Lyraes are unique and stand out
when compared to other objects. There are many advantages to classifying RR Lyraes, as they can be used as
standard candles and probes of the gravity field in the outer halo of our galaxy. Previous work to identify RR Lyraes
has only given a binary yes or no result, rather than a quantitative measurement. This project developed a
process to determine the probability that an object is an RR Lyrae by using the random forest machine learning
algorithm. Random forest, implemented in the Python programming language, takes a training sample and creates
a classifier. This classifier can then quantitatively judge whether other light curves resemble RR Lyraes.
Probabilities were successfully generated for a sample of 1257 potential RR Lyraes. Future work could incorporate
more features such as corrections for weather, color, etc., as well as completeness corrections.
Oxygen Isotope Studies of Mineral Separates in Eucrite Meteorites
Andrew Heard
Mentors: Edward Stolper and Maryjo Brounce
The eucrites are basaltic meteorites belonging to the howardite, eucrite, and diogenite (HED) clan. The petrology
and oxygen isotope compositions of the HEDs suggest that they derive from a common, differentiated parent body,
likely asteroid 4 Vesta. Experimental work has demonstrated that eucrite compositions cluster around the 1atmosphere olivine, pyroxene, plagioclase peritectic, leading to the hypothesis that eucrite meteorites represent
low-degree partial melts from a differentiated source. However, 26Mg excesses in HEDs indicate that the young
planetary body was hot, due to decay of the extinct radionuclide 26Al. Heating by 26Al-26Mg decay may necessitate
high-degree, rather than low-degree, partial melting. There has been significant work documenting HED whole rock
oxygen isotope compositions, however separate mineral phases have received less attention. Oxygen isotope
partitioning between minerals is temperature dependent, enabling a detailed study of thermal histories and
potential means to discriminate between petrogenetic models for eucrite formation. Preliminary modelling results
using parameterised partition functions, mass balance, and available published data, suggest that oxygen isotope
fractionations between pyroxene and plagioclase reflect temperatures ranges of 500-1100OC. Additional high
precision laser fluorination analysis of mineral phases will illuminate the thermal history of eucrite meteorites. Data
will be collected pending resolution of equipment maintenance issues.
Effects of the Rocky Mountains on Precipitation in Eastern North America
Alexander Henny
Mentors: Simona Bordoni and Jinqiang Chen
This paper analyzes the atmospheric flows around and east of the Rocky Mountains in the context of the moist
static energy (MSE) budget. Using ERA-Interim Reanalysis data, we decompose the horizontal moist enthalpy
advection, the horizontal dry enthalpy advection, and the latent energy advection into mean, transient, and
stationary eddy fluxes. These terms are interpreted in the context of orography and known atmospheric processes,
in order to establish why the Rockies do not induce a monsoon similar to that seen in East Asia.
Characterizing the Elemental Compositions and Temperature Structures of Brown Dwarfs and
Exoplanets
Erich Herzig
Mentors: Heather Knutson and Bjrn Benneke
Spectroscopic observations of exoplanets and brown dwarfs contain crucial information about their atmospheric
composition and temperature structures. In this work, I use the newly developed atmospheric retrieval framework
SCARLET to recover the atmospheric properties of these objects from their thermal spectra. The framework is
tested on the brown dwarf Gl570D with plans to study the atmospheres of other brown dwarfs and giant
exoplanets. The results produced by running these retrievals will crucial constraints on the formation histories of
giant planets and brown dwarfs as well as the chemical and physical processes in their atmospheres.
Developing an Automated Process for Quantifying Cellular Structures in Hydrogen Flames
Morgan Hill
Mentors: Guillame Blanquart and Jason Schlup
Our current simulations of lean hydrogen combustion show cellular instabilities, which are computationally taxing
to represent. As lean hydrogen mixtures combust, the flames develop cell-like instabilities, which can potentially
lead to flame extinction or blowback, where to the flame goes out or escapes to the fuel source, respectively.
Understanding hydrogen flame instabilities more completely will enable these processes to be mitigated by
performing accurate and computationally efficient simulations to aid the design of new burners. The lab's current
simulations with hydrogen mixtures are performed at atmospheric pressure with lean equivalence ratios, indicating

30

a deficit of hydrogen relative to the amount that would be needed to fully combust in air. In this work, an
automated program is developed to validate the hydrogen combustion simulations. Using multiple experimental
data sets collected from other researchers, an image-processing algorithm is designed and tested to both sharpen
the images and detect edges between the cellular instabilities. Multiple edge detection and sharpening schemes
were utilized before selecting a best performing set of parameters. Finally, once the images are processed
sufficiently, cell size and shape are determined and catalogued.
Design and Growth of Broadband Anti-Reflective Coatings for III-V Photoelectrochemical Devices
Alec Ho
Mentors: Nate Lewis, Erik Verlage, and Stefan Omelchenko
Photoelectrochemical (PEC) devices use solar radiation to split water into molecular hydrogen and molecular
oxygen for use as chemical fuels. High-efficiency III-V semiconductors such as GaAs have ideal band gaps that
make them desirable for PEC cells, but are unstable in 1M KOH. TiO2 functions as a viable protection layer that also
acts as a single-layer antireflective (AR) coating. However, a thick layer of TiO2 is needed for sufficient protection,
and therefore a single-layer AR coating is non-ideal. Adding a second layer gives design flexibility to reduce
reflective losses. The focus for this research is to increase the efficiency of multi-junction PEC devices by designing
an optimal double-layer coating without altering the chemical stability of the oxide surface. In creating the doublelayer AR coating, transparent thin films were deposited by atomic-layer deposition and radio-frequency sputtering
onto GaAs and optically characterized by ellipsometry and electrochemically characterized by cyclic voltammetry
and chronoamperometry. Stable samples with low refractive indices were patterned with photolithography.
Thicknesses were optimized for the AM 1.5 spectrum for the ideal 1.8/1.2 eV band gaps and deposited by rfsputtering followed by electrodeposition of nickel. SiO2 and Y2O3 are viable candidates for integration with TiO2 in
record efficiency PEC devices.
Bulk Friction Angles in Dry, Drained, and Saturated Gravel Beds
Samuel Holo
Mentors: Michael Lamb and Marisa Palucis
We examined the effect of capillary action and lubrication of grains on bulk friction angles through tilting chute
experiments. In each experiment, we screed a bed of 5mm gravels in a 65cm x 18cm tilting chute and slowly tilted
the chute until a granular avalanche occurred. We performed these experiments with a bed of dry grains and with a
bed that had been submerged and subsequently drained such that no water occupied the pore space. For each of
these cases, we performed experiments with 5, 10, and 15cm bed thicknesses. In the dry case, the bed failed at
~ 41, and bed thickness did not have a significant effect on failure angle. In the drained case, friction angles
increased from 46.5 to 50.9 with increasing bed thickness. The observed increase in friction angles between the
dry and drained cases suggests that addition of the water induces some sort of cohesive effect on the grains and
that this effect increases with bed thickness. Ongoing work includes experiments with fully saturated beds to
differentiate between capillary and lubrication effects, as well as debris flow initiation modeling at the catchment
scale.
Role of Retinoic Acid Pathway on Zebrafish Enteric Nervous System Development
Stephanie Hong
Mentors: Marianne Bronner and Rosa Uribe
The enteric nervous system (ENS) controls the peristaltic movement of food through the gastrointestinal tract.
During development, it derives from vagal neural crest cells (VNCC) that migrate into the developing gut to
become a diverse class of neurons and glial cells. In avian and mouse embryos, the Retinoic Acid (RA) signaling
pathway has been shown to induce migration and population growth of VNCC, improving the ability of VNCC to
colonize the gut in chains. To date, there are no published functional roles for the RA pathway during zebrafish ENS
development, despite its strong expression therein. Previously, our preliminary experiments showed that RA
application prior to 48 hours post-fertilization (hpf) induced a greater number of NCC in the gut region than
control, eventually resulting in increased enteric neuron number later in development. In order to further
investigate these observations and test the hypothesis that the RA pathway plays an important functional role
during ENS development, the focus has been shifted from gain of function effects of the RA pathway to the effects
of loss of function. The transgenic fish line Tg(hsp70:dn-zrar-gfp) expresses a dominant-negative Retinoic Acid
receptor, Raraa, fused with a GFP molecule on its C-terminal domain, the expression of which is controlled by the
heat shock promoter hsp70 such that incubation of embryos at 38C results in its expression. Expression of this
dominant negative construct attenuates RA signaling in a temporally controlled manner. Zebrafish embryos were
heat shocked at 24 hours post fertilization (hpf) and assayed by in situ hybridization and immunohistochemistry for
genes that mark neural crest cells or the gut environment. RA attenuation resulted in expansion of the RA pathway
component aldh1a2 mRNA expression and diminishment of meis3, a downstream RA target. Furthermore, heat
shocked embryos exhibited morphed and irregular patterning in the pharyngeal arches, of which NCCs are a
precursor, compared to the controls at 72 hpf. Embryos were also double-heat shocked to increase the severity of
RA pathway loss of function. Using phox2bb as a marker for migrating NCC in the gut, it was found that a one-time
heat shock caused NCCs to migrate farther down the gut but in lower numbers, whereas the NCC migration of

31

double-heat shocked embryos was delayed. However, when crestin was used as a marker for all neural crest,
single-heat shocked embryos displayed delayed migration compared to the controls. These loss of function results
indicate that the RA pathway plays a complicated role in influencing VNCC migration into the developing gut and
suggest that RA is a highly conserved signaling pathway that VNCC utilize in order to colonize the developing gut.
Engineering Conditional CRISPR Interference
Andrew Hou
Mentors: Niles Pierce and Mikhail Hanewich-Hollatz
CRISPR/Cas9 is a novel gene editing tool which has been repurposed for regulation of gene expression, via a guide
RNA which confers sequence-specificity and a catalytically dead Cas9 endonuclease (dCas9, for short). Termed
CRISPR interference (CRISPRi), dCas9 and the guide RNA form a complex which binds to and blocks transcription
of a specific target gene. Using RNA design principles, we sought to re-engineer guide RNAs such that CRISPRi can
be conditionally activated or inactivated by synthetic or endogenous inputs in E. coli. To this end, we demonstrated
that guide RNAs can be inactivated either by binding a separately expressed antisense strand, or by selfcomplementarity. However, as is, the guide RNA sequence is heavily constrained. In order to obtain the
programmability required for conditionality, we added extra nucleotides to the loop region found in the secondary
structure of the guide RNA. We hypothesize that: (1) a guide RNA with an extended loop could be conditionally
inactivated by an input strand which is antisense to the loop, and (2) a guide RNA with self-complementarity to the
extended loop would be natively inactivated, and that it could be conditionally activated via toehold-mediated
strand displacement.
Hardness vs. Randomness Tradeoffs for Arthur-Merlin Games in the Uniform, Extreme Low-End Setting
William Hoza
Mentor: Chris Umans
We investigate candidates for a polynomial-stretch hitting set generator against uniform conondeterministic
circuits, with an associated commit and evaluate protocol for establishing correctness. Such a hitting set generator
would imply subexponential-time nondeterministic simulations of AM in the uniform average-case setting, under
the assumption that E is not contained in AM.
The Role and Expression of Myc-Nick in Migratory Neural Crest Cells
Jenny Hsin
Mentors: Marianne Bronner and Laura Kerosuo
The neural crest is a transient population of multipotent stem cells that form in the dorsal part of the developing
neural tube. As the neural tube closes, the neural crest cells go through an epithelial to mesenchymal transition
(EMT) that allows them to migrate out of the neural tube and into various target tissues in the developing embryo.
Two members of the Myc transcription family, c-Myc and n-Myc, are expressed during neural crest development. In
addition to Mycs nuclear, DNA binding function, Myc can also be post-translationally cleaved into a cytoplasmic
form called Myc-Nick that recruits acetylases to microtubules and therefore causes cytoskeletal rearrangement. So
far, Myc-Nick expression has not been shown in vivo and the goal of this project was to examine whether neural
crest cells express Myc-Nick. We also wanted to compare the expression pattern of n-Myc and c-Myc in the neural
crest cells. For this approach, we cloned Flag-tagged versions of full length c-Myc and n-Myc; truncated Myc-Nick;
and a mutated c-Myc called c-Myc 295, which cannot form Myc-Nick into a chicken expression vector. We used in
ovo electroporation and immunostaining to verify whether the proteins were located either in the nucleus or the
cytoplasm at various stages of neural crest development. We also used double fluorescent in situ hybridization
(FISH) to efficiently compare the expression differences of n-Myc and c-Myc mRNA. Our preliminary results suggest
that nuclear Myc is expressed in the neural tube prior to EMT and Myc-nick is formed in the migrating neural crest
cell population. At the mRNA level, in premigratory stages, expression of n-Myc is dominant in the neural portion of
the neural tube whereas c-Myc is only expressed in the dorsal neural tube where the neural crest is formed.
However, both Myc family members are expressed in the migrating neural crest population after EMT. Our results
suggest interesting new and separate roles for the Myc family members in the premigratory versus the migrating
neural crest cell populations and function experiments to reveal this are ongoing.
Time Evolution of a Damped Quantum Oscillator
Tianyi Hu
Mentor: Vidar Gudmundsson
We consider the time-evolution of a damped quantum oscillator. The oscillator is weakly coupled toa thermal
reservoir of high temperature. The Liouville-von Neumann equation of motion for the density operator is
transformed into a master equation for the reduced density operator by tracing out the varibles of the reservior. As
the system is Stark-shifted by the potential V0 (t) (x/a), where is the Heaviside step function and a is the
characteristic length of the harmonic oscillator, we can solve the time-dependent master equation by two different
methods. In the first method, we use direct iteration via the Cranck-Nicholson method in the Hilbert space made
from the eigenstates of the harmonic oscillator. We then calculate mean values as the trace over the density matrix

32

and the corresponding operator. In the second method, we first seek the steady state solution in the Liouville space
and then use transformations to derive the time-evoultion without the need to numerically intergrate the equation.
With these two methods, we can analyze the complex spectrum of the Liouville operator with a harmonic or
anharmonic oscillator. In addition, the pumped quantum oscillator is briefly examined. We use Fortran 2008 to
achieve fast parallel programs with high flexibility. The aim of the project is to test methods on a simple open
quantum system that will be used on a more complex model of interacting electrons transported through a photon
cavity.
Stress-Induced Alterations in Behavior Are Mediated by the Tac2 System
May Hui
Mentors: David Anderson and Moriel Zelikowsky
Stress is a behavioral response found in virtually all animal species. Recently, Tachykinin 2 (Tac2), which encodes
the neuropeptide Neurokinin B (NKB), has been implicated in fear memory consolidation. We were interested in
exploring the possibility of a larger role for Tac2 in mediating stress. In particular, we examined whether systemic
and local antagonism of the NKB-specific receptor, NK3, with the drug osanetant, could alter the effects of various
manipulations to induce stress. Mice were subjected to chronic stress comprised of either social isolation, restraint
stress, or footshock, given daily across 14 days. Mice were then tested for stress-induced changes in behavior
using a looming assay (as a measure of innate defensive behavior) and trace fear conditioning (as a measure of
cognitive function and stimulus reactivity). At test, mice were treated with a systemic injection (i.p.) of osanetant
or microinfusions into the dorsal-lateral striatum or the dorsal bed nucleus of the stria terminalis. We found that
osanetant was able to buffer against the effects of chronic stress. Moreover, we found that systemic administration
of the NK3 agonist senktide was able to induce stress-like effects in non-stressed, group-housed control mice.
Collectively, our data point to a broad role for Tac2 in the regulation of chronic stress.
Bayesian Updating of Uncertain Parameter Vectors by Sampling From an Auxiliary Dynamic System
Hanna Hultin
Mentors: James Beck and Thomas Catanach
Bayesian updating can be used to learn uncertain model parameter vectors with sensor data from a dynamic
system. A fully probabilistic Bayesian model updating approach requires the evaluation of multidimensional
integrals which usually cannot be done analytically. Instead sampling based numerical methods can be used,
although high-dimensional problems are still challenging for current sampling methods. The aim of this project is to
examine how to use theory from stochastic dynamics to generate samples from a high-dimensional distribution
efficiently. This method, which we call the Langevin Monte Carlo method, should work well in higher dimensions
because the samples naturally move to areas of high probability according to the stochastic dynamics. To reduce
the unproductive computational effort during the settling-in period, as short settling time and low correlation as
possible is sought. In this project, the method used to reduce the settling time is by choosing the damping and
temperature schedule for a virtual annealing scheme.
Modelling Anisotropic Particle Distributions for Solar Probe Plus
Olivia Humes
Mentors: Mark Wiedenbeck and Richard Leske
The Solar Probe Plus mission will investigate the dynamics of solar energetic particles using the High Energy
Telescope and Low Energy Telescopes. Monte Carlo simulations were used in order to model the angular response
of these particle detectors. Typical inferred best fit values were calculated from simulated detector counts for
various combinations of parameters: magnetic field zenith angle and azimuth angle, as well as values of the first
and second order anisotropies. These data will be useful for understanding the responses of the High and Low
Energy Telescopes once they begin taking measurements for the Solar Probe Plus mission.
Spin Selectivity of Electron Transmission Through Solvated Double-Stranded DNA
Sylvia Hrlimann
Mentors: Jacqueline K. Barton and Theodore Zwang
In recent years, double stranded DNA (dsDNA) has been shown to have a large spin polarization in electron
transmission that is several orders of magnitude larger than what the current theory predicts. A study of how
tertiary structure affect the spin polarization of dsDNA could shed light to the biological implications of the effect
and help improve the theory. Thiol labeled dsDNA was attached to gold capped nickel substrates in order to inject a
current in the dsDNA and vary its spin polarization. Progress has been made towards attaching the dsDNA to nickel
and investigating the effect that mismatches in dsDNA has on spin selectivity.

33

Engineering and Understanding Light Sensitive Proteins With Applications in Live-Cell Imaging
Nicholas Hutchins
Mentors: William Clemons, Jr., and Austin J. Rice
Red-fluorescent proteins (RFPs) provide an alternative to the typical blue and green fluorescent proteins used in
cellular biology. The red light used to visualize RFPs is less cytotoxic, scatters less, and avoids natural cellular
autofluorescence. However, RFPs have two disadvantages: they are typically tetrameric in their native state and
they are dimmer than their green and blue counterparts. We show that RFP monomers can be created from a
natively dimeric RFP without substantial loss of fluorescence using computational design coupled with directed
evolution. To better understand the molecular basis of our improved RFPs, we pursue structural information
through protein crystallography.
Creating a Platform Independent GUI Utilizing Bayesian Rapid Optimal Adaptive Design (BROAD) to
Study Time Preference and Risky Choices
Seong Bin Im
Mentors: Colin F. Camerer and Romann M. Weber
Many different models have been proposed to account for time-inconsistent preferences and risky behaviors in the
field of behavioral economics. Experimental comparison is essential to testing these models and their implications.
However, most of the standard designs tackling such problems are not practical due intractability of these models
in large design spaces. Bayesian Rapid Optimal Adaptive Design (BROAD) takes advantage of adaptively optimizing
questions based on previous response of each iteration and thus allows for fast, real-time implementation with
performance guarantees. In order to make such method more wide-spread and user-friendly, we create a platform
independent Graphical User Interface (GUI) utilizing BROAD. As a prototype, we aim to build the interface for risky
choices, such that we may generalize to fit other areas of interest such as time preference.
Engineering Chimeric Antigen Receptor T-Cells for the Treatment of HIV
Erin Isaza
Mentors: Pamela Bjorkman and Rachel Galimidi
Chimeric antigen receptors (CARs) are engineered receptors that graft the specificity of a monoclonal antibody onto
a T-cellthe cells are removed from a patient, modified to express the desired receptors via retroviral vectors, and
reintroduced into the patient. While the only previous attempt to use CAR T-cells to attack HIV proved
unsuccessful, in the time since this study broadly neutralizing anti-HIV antibodies, which are able to neutralize
many strains of HIV, have been discovered. We have engineered CAR T-cells expressing broadly neutralizing
antibodies, which bind to and neutralize HIV with a much higher avidity and are safer than the first generation of
anti-HIV CAR T-cells.
Improved State Estimation and Control of a Pioneer 3-DX for a Resilient Software Executive
Riashat Islam
Mentor: Richard M. Murray and Catharine L.R. McGhan
Autonomous navigation systems that can operate safely without collisions in an unknown environment with little to
no human intervention can have significant applications towards space exploration missions for science discoveries.
In this project, we strive to further develop the initial implementation of a resilient risk-aware software architecture
that can handle uncertainty in hazardous environments. We present a framework that integrates real time planning
capabilities for an autonomous robot that uses the observed maps from the environment for planning and obstacle
avoidance algorithms. We propose a model that uses the Robot Operating Software (ROS) for real-time
environment map generation using Hokuyo lidar sensors, and use generated maps for navigating and avoiding
obstacles in an unknown environment. Furthermore, we work towards integrating simultaneous mapping and
localization (SLAM) techniques for the autonomous vehicle to be able to localize itself in the unknown generated
environment map. Our results suggest the integration of real time mapping capabilities into the resilient software
architecture, using Hokuyo lidar sensors mounted on top of the autonomous vehicle, on both simulated and
practical environments. We demonstrate results of implementation of obstacle avoidance algorithms, along with
existing probabilistic planning algorithms, for safe navigation using the maps generated by the laser sensors. We
also demonstrate that SLAM based techniques can be integrated into the resilient software architecture of the
autonomous vehicle.
Encapsulins as Molecular Reporters for Spatiotemporal Control of Biological Systems
Vasant Iyer
Mentors: Mikhail G. Shapiro and Pradeep Ramesh
Continued advancement in molecular medicine and our understanding of biological systems depends on developing
techniques to modulate gene expression, cell behavior, and tissue properties in a spatiotemporally specific and
noninvasive manner. We aim to develop such a technique using proteins of a new class of bacterial nanoparticles
called encapsulins, which natively contain guest proteins that enable bacteria to sequester iron in times of oxidative

34

stress. We have targeted proteins to the interior of encapsulins to create a redox-controlled environment in which
magnetite particles may be nucleated upon iron supplementation, yielding T2 contrast under magnetic resonance
imaging (MRI). Further work involves developing encapsulins as noninvasive agents for cellular control based on
hyperthermia induced by incident radio frequency waves.
Extreme Scale Parameter Discovery Using Randomized Metric Embeddings
Nauman Javed
Mentors: Sumit Kumar Jha and Axel Mller
In many fields, complex computational models can exist in high dimensional parameter spaces in which
conventional optimization methods are computationally infeasible. A significant reduction in the search space
dimension combined with the intelligent use of probabilistic metaheuristics can drastically reduce the computational
cost of many optimization problems, and would pave the way for increasingly sophisticated and informative
models. We introduce a novel dimension reduction and search algorithm that logarithmically reduces a given
models parameter space dimension via the Johnson Lindenstrauss lemma and searches for the model optimum in
the lower dimensional space using simulated annealing. Once located, the global optimum is reconstructed in the
original higher dimensional space by exploiting the structure preserving properties of the JL transform and simple
distance geometry arguments. We implement our algorithm in CUDA C to parallelize computation and experiment
with a range of computational models and initial parameter space dimensions. Our results demonstrate that the
algorithm can successfully approximate the global optimum of the parameter discovery problem for a sufficiently
large initial dimension. Furthermore, we demonstrate that the algorithm significantly reduces the computational
cost of model optimization.
Arterial Wall Shear Stress Imaging of Flow Phantom Using 4D MRI
Cynthia Jiang
Mentors: Albert Hsiao, Geno Pawlak, and Steven Low
Magnetic resonance imaging (MRI) is an important modality used in the diagnosis and treatment of cardiovascular
diseases, whether congenital or acquired. Until recently, planar two-dimensional MRI techniques have been
predominantly used for visualization of the heart. However, four-dimensional Flow MRI, a specialized MRI sequence
that allows for measurement and visualization of time-resolved three-dimensional blood flow patterns with velocity
encoding, is now steadily becoming more widespread as a more comprehensive, and time-efficient method of
obtaining flow and anatomic data.
Early work with 4D Flow MRI has proposed that hemodynamic flow patterns and wall shear stress may contribute
to the formation and growth of ascending aortic aneurysms. Measuring and calculating the wall shear stress may
be predictive of which patients eventually require surgery to repair these aneuryms. Our objective is to determine
the accuracy of calculated fluid shear stress and other hemodynamic parameters from 4D Flow MRI, given
limitations in its spatial and temporal resolution when used for routine clinical scanning.
Traveling Wave and Fractal Analysis of Complex Waveforms in Marangoni Driven Systems
Katherine Jiang
Mentor: Sandra M. Troian
The rapid spreading of an insoluble surfactant monolayer along the surface of a thin liquid film of higher surface
tension is known to be driven by Marangoni stresses. These forces, which pull the monolayer and underlying liquid
toward regions of higher surface tension, are proportional to the local magnitude of the surface tension gradient
caused by the local concentration gradient. Previous experiments in our lab using liquid films of the order of 10
microns in thickness have verified the propagation of an outer stable circular ring behind which there often
develops a bifurcating fractal front. While theoretical models predict that the ring should advance in time as t1/4, we
consistently find higher exponents ranging from about 2/5 to 1/2. Image analysis to determine the fractal
dimension of the trailer bifurcating front, using various thresholding and edge detection techniques, has revealed
differences from the values obtained in conventional hydrodynamic systems. In contrast to the well-known fractal
exponent of 1.67 associated with the Saffman-Taylor instability, we find values closer to 1.85. This significant
difference may help elucidate the physical mechanism responsible for the fingering instability observed.
Developing Temperature-Sensitive Proteins From a Bacterial Coiled-Coil for Thermal Regulation of
Genes
Yiwei Jiang
Mentors: Mikhail Shapiro and Dan Piraner
Thermosensitive proteins may enable selective regulation of genes with an external heat source, providing a
noninvasive technique for studying gene regulation. The coiled-coil motif of the TlpA protein allows it to sense
temperature shifts directly and modulate the extent of transcriptional repression by blocking its own promoter. We
are harnessing this system to demonstrate temperature-dependent gene up and down-regulation in both
prokaryotic and eukaryotic cells. In the most straightforward implementation of this system, we are expressing an

35

orange fluorescent protein gated directly by TlpA, resulting in protein production above a critical dissociation
temperature. In another construct, TlpA blocks the expression of the lacI repressor, thereby allowing a Lac
operator-gated promoter to drive expression of a red fluorescent protein at low temperature. At higher
temperatures, the TlpA uncoils, enabling transcription of lacI repressors and down-regulating red fluorescence.
Future experiments will use fluorescence measurements to observe brightness of cells transformed with the RFP
plasmid co-plated with cells transformed with the orange fluorescent protein. The mammalian implementation of
the TlpA system uses clustered, regularly interspaced, short palindromic repeat (CRISPR) technology to generate
RNA-guided nucleases, like Cas9, with customizable specificities for genome editing. Alternatively, catalytically
inactive mutants of Cas9 called dCas9 can be paired with short-guide RNAs (sgRNA) and block transcription. We
are currently validating a previously published system in which dCas9 tethered to the VP64 transactivator enhances
expression from a minimal CMV promoter with the appropriate upstream sgRNA binding site. Future work involves
replacing upstream tandem Lac operator with the TlpA operator to enhance fluorescence above the TlpA critical
temperature.
Waypoint Sequencing in a Dynamically Shadowed Environment
Jonathan Joo
Mentors: William Red Whittaker and Antonio Rangel
Future missions to explore planetary surfaces depend on the ability of solar-powered rovers to visit multiple
waypoints in shadowy environments. The order in which these locations are visited, called the waypoint sequence,
is crucial to the overall success of a mission. While efficient routes maximize the net value of visited locations,
inefficient routes may waste time navigating to dead-end regions or re-routing to previously visited areas. Current
studies have investigated efficient travel from one point to another in such an environment, but there does not yet
exist an effective way to plan for a multiple-location mission. Furthermore, existing approaches to similar
sequencing problems have never been applied to a time-variant shadowy environment such as that of the lunar
poles. This paper thus describes a genetic algorithm to create efficient waypoint sequences in dynamically lit
regions. The genetic algorithm greatly outperforms a brute force approach in calculation time, while generating
near-optimal paths. Furthermore, this algorithm obtains significant improvements in value optimization compared
to a simple greedy algorithm. Thus, this research demonstrates that a genetic approach could be utilized to
effectively plan future missions for solar-powered rovers in dynamic, shadowed environments.
Discovering RNA Structural Motifs Responsible for lncRNA-Protein Interactions
Rushikesh Joshi
Mentors: Mitchell Guttman and Mario Blanco
It is now apparent that our genome is primarily comprised of non-coding RNAs (ncRNAs), and not protein-coding
messenger RNAs. Advances in high throughput RNA sequencing have established the existence of thousands of
large non-coding RNAs (lncRNAs) that are involved in diverse biological and regulatory roles.
Similar to classical ncRNAs, the function of lncRNAs depends on RNA-protein interactions. RNA structure
corresponds directly to both protein-binding affinity and function. We know that lncRNAs act as flexible, modular
scaffolds, and are essential for assembling protein complexes based on highly specific, discrete, protein-binding
domains located within the lncRNA. We aim to characterize the RNA secondary structure motifs that are present at
these protein-binding domains.
By combining high throughput sequencing techniques and in vivo application of RNA structure probing reagents,
which react at the 2-hydroxyl group of conformationally dynamic RNA nucleotides, we can determine regions of
structural significance. We have been able to utilize computational methods to identify and characterize the
structure of known RNAs by comparing our results to previously determined two-dimensional structures.
We plan to apply our methods on a genome-wide scale, to characterize the structure of lncRNAs. This will help us
understand the structural motifs responsible for lncRNA-protein complex formation.
Predicting the Behavior of PAMAM Dendrimers in a Water, Methanol, and Sulfamethoxazole Solution
Using Molecular Dynamics Simulations
Sachi Kamiya
Mentor: Jose Mendoza Cortes
Molecular dynamics from LAMMPS is a powerful tool to predict the atomistic behavior of materials/molecules in
different environments. Additionally, the study of biocompatible materials has become a growing field in the
sciencesit has become especially relevant to understand how foreign materials react and behave in the body. In
this project, LAMMPS, a molecular dynamics simulation tool, is used to simulate the behavior of dendrimers, a
widely studied biocompatible molecule, to understand its atomistic behavior. In order to set up the experiment,
generation 3 and generation 4 polyamidoamine (PAMAM) dendrimers are placed in different simulations including a
water, methanol, and drug solvent. The simulations are run in an NVT (canonical system) and NPT (isobaric
system) where the system is heat up to 300K and pressure is kept constant to 1atm. The output data is then

36

analyzed using the radial distribution function (RDF) and velocity auto-correlation function (VAF) to understand the
intermolecular behavior of the dendrimer and surrounding environment. To study spontaneity, the free energy of
each system is also evaluated. Further potential research areas includes the potential of studying various other
drugs using explicit solvent.
Spectral Studies of Seyfert 2 Active Galactic Nuclei From the NuSTAR Hard X-ray Observatory
Nikita Kamraj
Mentors: Fiona Harrison and Liz Rivers
Active Galactic Nuclei (AGN) are some of the most luminous compact objects in the universe, frequently outshining
their host galaxies. Unification schemes propose that all AGN are essentially the same physical object, with the
diversity of AGN classes simply arising from different orientations of the AGN with respect to the observer. This
project focuses on the study of the X-ray spectral features of Seyfert 2 AGN observed with the NuSTAR telescope.
The high energy X-ray focusing optics of the NuSTAR satellite have enabled AGN spectral features to be resolved to
unprecedented detail, allowing tighter constraints to be placed on the geometry of the circumnuclear material. I
performed data reduction and extraction processes using the NuSTAR Data Analysis software, followed by spectral
modeling of the data using the XSPEC fitting tool. My investigation has focused on modeling the spectra of NGC
4388, NGC 6300, NGC 7172, NGC 1052 and IC 5063, with the results interpreted in the context of Seyfert 1/2
unification schemes.
Spatiotemporal Online Imitation Learning
Andrew Kang
Mentors: Yisong Yue and Hoang Le
Spatiotemporal sequence prediction is a problem of increasing interest within machine learning. The example of our
focus is imitation learning for online prediction of camera planning, in which the goal is to mimic a human camera
operator. However, such learning is not feasible with standard supervised learning approaches as they do not learn
temporal patterns. Therefore, a promising alternate solution is the state-of-the-art approach that uses reductions
with a focus on non-linear predictors. This approach is best seen in two particular algorithms, SEARN and DAGGER.
However, two challenges arise in extending these algorithms to spatiotemporal sequences: each sequence can get
arbitrarily large, and we are no longer working with discrete values. Thus, the algorithms must be modified to
overcome these obstacles; namely, we focus on proving that the regret and loss bounds still hold when averaging
between policies in the continuous setting instead of sampling in the discrete setting. That is, in this paper, we
examine the theoretical guarantees we can make about an algorithm that provides minimal loss and good
generalization for the camera planning prediction problem.
Band Structure Measurement of Gyroid Photonic Crystals in the Mid-Infrared
Emil T. Khabiboulline
Mentors: Harry A. Atwater and Siying Peng
Like electronic topological insulators, photonic topological insulators form a new class of materials with exciting
properties that have implications for many areas, including photovoltaics. The topology of their band structure
gives rise to propagation of light that is unidirectional, fault-tolerant, polarized, and localized to the boundary.
Theory predicts that gyroid photonic crystals are photonic topological insulators. We aim to provide experimental
confirmation in the mid-infrared region.
Starting with two-photon lithography to fabricate the samples in polymer, a sequence of atomic layer deposition,
focused ion beam milling, plasma cleaning, and chemical vapor deposition lead to alumina structures coated with
amorphous silicon. Single and double gyroid crystals, with unit cells of 5 m organized in a 10x10x10 cube, were
fabricated this way on an intrinsic silicon substrate. We build a setup to perform angle-resolved spectroscopy using
a quantum cascade laser as the source of the ~8 m beam. From measurements of reflection and transmission
over a range of wavelengths and incidence angles, we construct the band structure of the photonic crystals. Theory
predicts band gaps for the single gyroid and Weyl points for the double gyroid. Comparison to experiment indicates
whether or not our nanofabricated gyroid structures possess a topological nature.
Engineering of Viral Vectors Selectively Expressing Cre Recombinase in Target Neural Cell Populations
of Mouse Model Expressing Human Mutant Huntingtin Gene
Hyun Min Kim
Mentors: Viviana Gradinaru and Benjamin Deverman
Adeno-associated virus vectors are commonly used gene delivery vehicles for basic science studies and are being
explored for human gene therapy. The viral capsid can be customized to create a recombinant virus expressing
packaged genes and possessing transduction characteristics that enable targeting of specific regions of the brain.
Cre-Recombination-based AAV Targeted Evolution (CREATE) has enabled the customization of AAV by significantly
expediting the selection of viruses possessing desired characteristics. CREATE is a Cre-dependent selection system
where capsids that transduce predesignated Cre+ target cell types are selectively recovered. This has optimized the
targeting efficiency of the AAV9, which has been demonstrated to cross the blood-brain barrier and deliver

37

transgenes to the CNS, resulting in the design of AAV-PHP.B. We are now enhancing the accuracy of the selection
process by modifying the mutant lox sites within the capsid genome used for CREATE. Furthermore, by using AAVPHP.B to deliver Cre regulated by promoters to neural cell populations, we are addressing questions regarding
neurological disease models. By using a Cre/lox-conditional mouse model expressing the human mutant huntingtin
gene (mHTT), we are seeking to identify the cell types that lead to Huntingtons disease when expressing mHTT
and quantify the fraction of mHTT expressing cells that must be suppressed to ameliorate symptoms.
Three Dimensional Image Recognition in Mice
Jaebin Kim
Mentor: Doris Tsao and Tomo Sato
By discovering the extent to which mice are capable of perceiving and recognizing a variety of occluded objects
surrounding them, and incorporating two-photon calcium imaging and electrophysiology to analyze the neuronal
circuits involved in visual perception, we can not only gain a deeper understanding of neuronal functions, but also
understand the mechanisms of visual perception that is conserved across species. We trained the mice to recognize
various visual stimuli and recorded their rate of success in choosing the correct stimulus for a juice reward. When
setups for two-photon calcium imaging and electrophysiology are installed, the mice will undergo the same
experiment but instead with the partially occluded visual stimuli. This will reveal if the mice are able to recognize
the occluded images, and indicate the neuronal activity and mechanism involved in three dimensional image
recognition.
Single Cell RNA Imaging on a Cellphone
Jinhyun Kim
Mentors: Long Cai and Ahmet F. Coskun
In information-oriented society, due to high-quality specifications, phones can be utilized as microscopic imagers
for biomedical applications. In this project, we demonstrate a compact home-built optical platform running on a
smartphone to image and quantify RNA molecules in single cells. This cellphone based RNA imaging platform is
hand-held and low-cost making it rather useful research toolset for scientists in resource poor settings. To show its
proof-of-concept, we devised an optomechanical attachment that was mounted on a Samsung Galaxy S4 camera
unit using a combination of 3D printed parts, an illumination source, an imaging lens, and an optical filter. We
optimized the performance of this approach using a USAF chart and fluorescent microspheres with 15-micron
diameter. For validation of transcript detection capability, we visualized highly expressed mRNA molecules in
mammalian cells using this portable cellphone RNA imager. This platform aims to democratize systems biology
particularly towards microbial diversity research.
Rotational Invariant Low-Rank Pose Estimation
Joon Sik Kim
Mentor: Yisong Yue
Estimating human pose from 2-D images allows to recover interesting information about the nature of human
poses. Our goal is to study how human poses change when performing different activities. To do so, we try to
recover basis of poses that summarizes the variability of poses in our dataset. The advantage of this representation
is that we can describe the many degrees of freedom of human natural poses in a compact way, using a relatively
small basis. 2-D images have the limitation that the same pose may be obtained from different 3-D poses
projected from different camera angles. In this work, we address this challenge with two possible approaches
based on a form of structured low-rank matrix factorization, (1) incorporate rotation invariance by learning a
sequence of 2-D pose bases (one for each angle of view), (2) try to recover directly the inherently invariant 3-D
pose basis. A detailed comparison of the two methods and analysis of the performances is carried out.
Optimization of Methane Monooxygenase to Convert Methane to Methanol on an Industrial Scale
Lani Kim
Mentors: Stephen Mayo and Alex Nisthal
Methane (CH4) is the second most prevalent greenhouse gas emitted in the United States, accounting for 10% of
all greenhouse gas emissions. By converting methane to methanol, the methane gas, instead of going to waste,
can be used to create biofuels and other higher value chemicals. Currently, the industrial methane-to-methanol
conversion process is extremely energy-intensive and environmentally unfriendly. Instead, the biological oxidation
of methane to methanol using the methane monooxygenase enzyme (MMO) is a more desirable conversion
process. By optimizing spmoB, an MMO variant, highly active methane-oxidizing enzymes can be created for
industrial use. In order to produce large amounts of spmoB, the variant is transformed into a Pichia yeast secretion
system. The effectiveness of the Pichia secretion system is then measured through absorbance and sGFP
fluorescence levels.

38

Genetic Exploration of Conserved Unknowns in C. elegans

Shawn Kim
Mentors: Paul Sternberg and Ravi Nath
Caenorhabditis elegans is a multicellular organism in the phylum of Nematoda. C. elegans is made of 959 somatic
cells, and has a complex developmental process, including embryogenesis, morphogenesis, and final growth to an
adult. Important developmental principals are conserved from C. elegans to humans. The simple nervous system of
C. elegans controls a great diversity of behaviors that can be easily quantifiable. C. elegans are estimated to have
over 20,000 distinct genes. Many of these genes are orthologous with human genes, meaning that they have
evolved from a common ancestral gene. Therefore, there is a high probability that orthologous genes have similar
functions. Of the multitude of orthologous C. elegans genes, the Sternberg Lab has identified 1000 orthologous
genes whose functions remain unknown. From this list, we identified 152 genes that are the only members of a
particular undefined gene family, which means they are less likely to be genetically redundant. We noted that 30 of
the 152 genes in an unknown gene family are found in neurons. Given C. elegans simple nervous system and
genetic flexibility, we can identify the function of a gene by comparing the behaviors between wild-type worms to
mutant worms with a null-allele. To look specifically at the effect of knocking out these genes, I have screened 19
genes of those 152 conserved unknown genes by applying tracking assay, pumping assay, avoidance assay, and
aldicarb assay to the strains, each of which has knocked out one gene.
Self-Assembly, Structure, and Properties of Brush Block Copolymer Nanocomposites
Shi En Kim
Mentors: Robert H. Grubbs and Alice Chang
Block copolymer nanocomposites are hybrid materials that demonstrate important applications as the combination
of different materials engenders novel properties not inherent in either component material. This introduces new
opportunities for control in their structure and property. We incorporate nanoadditives into a brush polystyreneblock-poly(ethylene oxide) (PS-b-PEO) or brush polystyrene-block-poly(lactic acid) (PS-b-PLA) matrix. As brush
block copolymers display greater rigidity and less chain entanglement than linear block copolymers, they are shown
to self-assemble into unique structures with large long-range domain sizes. The brush block copolymers are
synthesized by ring opening metathesis polymerization (ROMP). The nanoadditives we incorporate (linear
homopolymers, compact star polymers, and tantalum oxide nanoparticles) span different length scales and
functionalities. We report the self-assembled structure and optical properties of these brush block copolymer
nanocomposites, toward enabling the design of novel functional materials.
Targeting lncRNAs in Pluripotency Network Using the CRISPR-Cas9 System
Won Jun Kim
Mentor: Barbara Wold
Long non-coding RNAs (lncRNAs) are an emerging group of RNAs that do not code for proteins and yet act as
functional molecules themselves. An increasing number of lncRNAs are being identified to participate in cellular
events, such as transcriptional and translation regulation, chromatin organization, and reprogramming. Despite
significant research efforts to fish out novel functional lncRNAs, genetic perturbations targeting lncRNAs usually
yield little success: their tendency to localize within the nucleus limits the use of genetic perturbations targeting
cytoplasmic RNAs. Here we demonstrate a reliable method for gene knockout using the CRISPR-Cas9 system to
delete lncRNAs in mouse embryonic stem cells (mESCs). We choose 4930444M15Rik, one of the previously
identified lncRNAs that are activated during reprogramming of somatic cells. In order to test its functional role in
reprogramming and pluripotency, we study the transcriptome changes after 4930444M15Rik knockout in mESCs.
Our data suggest that the CRISPR-Cas9 system can efficiently and specifically delete the target lncRNA in mESC
genome as we successfully isolate heterozygous mutant cell lines (single knockouts). Homozygous mutant cell lines
(double knockouts) generated with this method reveal reduced viability, indicating a crucial role that
4930444M15Rik plays in mESC survival and proliferation. Further downstream cell-based assays (e.g. siRNAmediated gene knockdown) and high-throughput transcriptome profiling studies (e.g. RNA-seq) are needed to
better elucidate the exact function of 4930444M15Rik.
Neurotransmitters Modulate the Pulsation of the Jellyfish Cassiopea
Young Min Kim
Mentors: Paul Sternberg and Ravi Nath
Neurotransmitters and neuropeptides modulate neural circuits across the kingdom Metazoa. The effects of these
neural circuit modulators have been observed and characterized in model organisms, such as Caenorhabditis
elegans and Drosophila melanogaster. We investigated the effects of neuromodulators on Cnidaria, an earlybranching metazoan lineage; the last common ancestor between the phyla Cnidaria and humans existed around
600 million years ago. We chose to study the cnidarian Cassiopea, which is in the class Scyphozoa, commonly
known as the true jellyfish. Cassiopea, known as the upside-down jellyfish, are an ideal model organism for
behavioral research in early branching lineages because they sit on their bell, which allows counting of pulsation.
Taking advantage of our state-of-the-art jellyfish behavioral arenas, we can track Cassiopea pulsing behavior under

39

different stimuli for prolonged periods of time. We have tested numerous molecules on the jellyfish, including
melatonin, serotonin, glycine, dopamine, histamine, GABA (Gamma-aminobutryic acid), epinephrine, DPCPX(an
adenosine antagonist), and glutathione. Jellyfish treated with melatonin exhibited a quiescent behavior, with a
reduced pulse rate; likewise, serotonin also induced behavioral quiescence, suggesting an inhibitory role of these
molecules on the cnidarian nervous system. Glutathione induced an oscillation between fast and slow pulsation.
Furthermore, to better understand Cassiopea pulsation, we observed the effects of temperature or salinity on this
behavior. Increasing temperature directly increases pulsation behavior, and vice versa. On the other hand, while
the jellyfish did not show any consistent behavior in response to high salinity, they showed drastically decreased
pulsation in low salinity. Our data suggest that neuromodulation of Cassiopea provides this animal with the ability
to control its pulsation rate under different environmental stimuli, which in turn also implies that neuromodulation
of neural circuits has been conserved throughout the course of evolution.
Investigating BSCCO High Temperature Superconductors With Magnetometry and Annealing
Gillian Kopp
Mentors: Dan Dessau and Nai-Chang Yeh
Superconductors exhibit no electrical resistance and expel magnetic flux when cooled below a critical temperature,
Tc. High temperature superconductors can withstand higher magnetic fields than conventional superconductors,
expanding their potential applications, however, the theory behind their superconducting mechanism is not well
understood. My work focuses on lead-doped BSCCO samples, where the Tc depends on the oxygen content and
hole doping of the CuO2 layers, which are essential to their superconducting properties. Tc measurements are done
with a magnetometer and lock in amplifier a system that allows for measurement of the magnetic flux passing
through the cuprate sample. Tc dependence on doping and how annealing alters this has been investigated, and I
have determined that by annealing in vacuum, we are able to increase the Tc and keep the transition width narrow.
Additionally, a monochromater spectrometer as a diagnostic tool for a mid-IR laser used for ARPES is in progress,
and has involved setting up hardware, software, and optics.
X-ray Analysis of 7 Active Galactic Nuclei
Peter Kosec
Mentor: Fiona Harrison
We perform X-ray analysis of 7 AGNs using spectroscopic measurements from NUSTAR, SWIFT, XMM-Newton,
Chandra and Suzaku satellites. The sample was initially fitted with a range of models simulating an active nucleus
obscured by dusty molecular clouds, including Pexrav and MYTorus models. Subsequently, sources of interest were
chosen for further analysis. ESO 383-G018 shows unexpectedly low intensity in the hard X-ray band (above 30
keV) of its spectrum, where a steep cutoff is observed. This spectral feature can be fitted with a power law with
exponential cutoff at 19,3+8,5
5,5 keV, much lower than hundreds of keV observed in other AGNs. ESO 509-IG066 is a
galaxy pair and simultaneously an AGN pair based on the XMM data measured in 2004. However, only one source
can be fitted in NUSTAR and SWIFT spectra from 2014 to high statistical accuracy. Comparing with Chandra data
from 2011, we report that the second object has dimmed by a factor of 10 since 2004. Possible explanations
include fading of the nucleus by decreasing the mass accretion rate or obscuration of the source by a Comptonthick molecular cloud. Further investigation in the IR band is needed to distinguish between the possibilities.
Heat Treatment Stabilization of Yttria-Stabilized Zirconia Thermal Barrier Coatings
Giacomo Koszegi
Mentors: Katherine T. Faber and Matthew Johnson
Thermal barrier coatings (TBCs) are commonly used to protect the blades of gas turbine engines from high
operating temperatures that would otherwise result in deformation or melting. Yttria-stabilized zirconia (YSZ) is
particularly well-suited as a TBC due to its thermal conductivity and coefficient of thermal expansion. When YSZ
TBCs are prepared via plasma-spray physical vapor deposition, the resulting coating contains non-stoichiometric
and amorphous material, both of which compromise stability during thermal cycling. The goal of this research was
to use heat treatments at 350C and 700C for times ranging from 1-16 hours to correct stoichiometry and
promote crystallization. X-ray diffraction was used to evaluate the non-stoichiometric phases before and after
heating, as well as to provide a qualitative measure of amorphous content. Scanning electron microscopy was used
to visualize microstructure. Similar experiments could be conducted in the future using other TBC ceramics, such
as gadolinium zirconate.
Machine Learning for Fast Data Transfers at the Large Hadron Collider
Nikhil Krishnan
Mentors: Maria Spiropulu and Dorian Kcira
The Large Hadron Collider (LHC) is the most powerful particle collider in the world, and the Caltech Compact Muon
Solenoid (CMS) group produces massive amounts of data in its experiments at the LHC. As a result, the CMS group
has its own infrastructure for data transfers, implementing grid-based data analysis and global-scale networking.
Because the CMS experiment generates voluminous amounts of data, there needs to be a way to improve the

40

current data management tools to optimize speed of transfers and facilitate fast I/O access. Machine learning offers
numerous methods for such an optimization, as the field allows us to look at the present state of the network and
use information about recorded transfers to predict features about future data transfers, thereby reducing the
number of variables the data management tool has to consider. Using logged information from PhEDEx, the CMS
data transfer system, we created a large sample matrix of data transfers, partitioned it into testing and training
samples, and conducted Principal Component Analysis, a machine learning method that projects a large
multivariate dataset onto the two variables that most explain variance. We also worked on constructing a deep
learning neural network between various nodes of the CMS group, allowing the data management tool to learn by
itself over time and optimize data transfers based on completion time.
Development of a Control System to Stabilize Magnetic Fields Outside an nEDM Detector
Anita Kulkarni
Mentors: Bradley Filippone and Simon Slutsky
Experiments over the past several decades have attempted to measure a neutron electric dipole moment (nEDM)
with greater and greater sensitivity. The presence of a nonzero nEDM would lend support to potential explanations
of matter-antimatter asymmetry, among other ramifications. A highly uniform magnetic field is required to detect a
dipole moment to high precision since non-uniformities can create false signals through the geometric phase effect.
The uniformity of the field inside the detector can be improved by reducing the magnitudes of magnetic field
disturbances outside the detector. To do this, we program a system that controls the currents flowing to a network
of six large coils (connected in pairs) around the detector in three spatial directions. The currents are programmed
to generate magnetic fields that counteract external field disturbances in real time. The system implements
Proportional-Integral-Derivative (PID) control of currents based on magnetic field readings from eight probes near
the detector. Continuous improvements are being made in the systems performance, and finer control could be
obtained in the future by powering the six coils separately and modifying the control system accordingly.
Numerical Simulation of Phase Locked Synchronization in Coupled MEMS Oscillators With Applications
to Neurocomputing
Ankit Kumar
Mentors: Raj Mohanty and David Hsieh
The dynamics of linearly coupled, anisochronous, self-sustaining oscillators are numerically simulated, with the aim
of identifying regions of parameter space that exhibit phase locked synchronization and multistability. Robustness
of synchronization in the presence of varying levels of noise is studied. The two oscillator case is extended to a
linear chain and a two-dimensional grid of oscillators with nearest neighbor coupling. Pattern recognition is studied
by tracking the evolution of the network towards synchronized states while iterating the initial conditions and
coupling parameters. The experimental implementation of the network using self-sustaining MEMS oscillators could
lead to the practical realization of neurocomputers on a scalable architecture.
Starburst: Stellar Burst Observation Program Control Software Implementation
Lokbondo Kung
Mentors: Anthony Readhead, Jackie Villadsen, and James Lamb
The Starburst Project aims to implement the first facility dedicated to radio observation of stellar bursts using the
two 27-meter antennas located in Owens Valley Radio Observatory. Many of the projects custom built pieces of
equipment require custom built software to help facilitate operations. Therefore, the implementation of data
querying wrappers for embedded systems and data streaming routines for data pulling in the back-end of the
antenna pipelines was crucial for the success of the project. Research into the hardware and architectural design of
the software was heavily stressed to provide not only support for the project but possibilities of expansions or
replications of the system. The project also involved implementation of control system software to manage and
control the individual components at the front-end (receiver and related components). The new control system
aims to replace an outdated legacy control system. This project will ultimately provide custom Python software for
the Starburst Project that is mutable, testable, and documented.
Simulation and Characterization of Near-Subwavelength High-Refractive Index Contrast Gratings for
High Efficiency Spectrum-Splitting Multijunction Photovoltaics
Stephanie Kwan
Mentors: Harry A. Atwater and Sunita Darbe
High-refractive index contrast gratings (HCG) are two-dimensional arrays of high-refractive index dielectric
subwavelength-scale structures on the surface of a low-refractive index substrate. HCG have been shown to display
near-perfect reflection over 200 nm-wide tunable bands in the near IR range under idealized condition (e.g. for
normal incidence light of a given polarization). This remarkable phenomenon is attributed to Mie electromagnetic
resonances and has inspired research on applications of HCG in solar energy conversion. In spectrum-splitting
photovoltaics, optical elements divide white sunlight into multiple frequency bands, which are coupled into laterally
separated solar cells with bandgaps tuned to best convert the target band of light. In this project, HCG are

41

explored as long-pass filters for the Atwater Full Spectrum teams ultra-high efficiency spectrum-splitting
photovoltaic module. Rigorous Coupled-Wave Analysis simulations are used to identify geometric parameters for
TiO2, GaP, and a-Si HCG. In parallel, we experimentally measure the angle-dependent reflectance properties of the
gratings over the visible and near IR spectrum. The long-term goal is to identify and fabricate seven high-efficiency
HCG that can replace the costly and angle-sensitive Bragg reflectors currently implemented in the submodule
prototype.
Progress Toward the Synthesis of Dimethyl Oxeatamide A: Synthesis of Racemic tert-ButylPyOx
Netgie Laguerre
Mentors: Brian M. Stoltz and Samantha Shockley
Diterpenes are a large class of natural products that display a range of biological activities that cause them to be
potential pharmaceutical agents. However, the structural complexities of these natural products that commonly
include all-carbon quaternary stereocenters make them formidable targets for synthetic chemists. The Stoltz group
has recently developed the innovative methodology of palladium-catalyzed conjugate addition to create all-carbon
quaternary stereocenters in high enantiopurity under atmospheric conditions while employing a relatively
inexpensive catalyst. We propose to undertake the asymmetric synthesis of the diterpene, dimethyl oxeatamide A,
utilizing palladium-catalyzed conjugate addition.
The requisite ligand for palladium-catalyzed conjugate addition chemistry, tert-ButylPyOx, requires tert-leucinol as
a key intermediate in its synthesis. In order to determine the enantiomeric excess of the proposed conjugate
addition reaction, racemic tert-butylPyOx is required. Since no efficient preparative methods currently exist for the
racemic synthesis of tert-leucinol, we are developing a novel synthetic route to this intermediate and thus racemic
tert-butylPyOx this summer.
A Numerical Analysis of Spin-Spin Coupling in Multi-Body Systems
Frederick William Lake
Mentor: Konstantin Batygin
Many bodies in the solar system have rotation times synchronous with their orbital periods. However, a
phenomenon known as spin-orbit coupling, where a deformed moon is gravitationally torqued by its central body,
permits alterations in the evolutionary process of gravitationally bound bodies, and has been fairly well studied at
this point. This paper analyzes the effects of a second, related phenomenon known as spin-spin coupling, in which
the spin of a central body affects the rotation rate of the secondary. A numerical integration program was written,
and applied to multi-body systems such as 87 Sylvia and 216 Kleopatra to show the effects of spin-spin coupling on
the systems. This program was then released publicly, as no such program is currently available for public use,
allowing further study into the phenomenon.
Development of a Three Wavelength Cavity Ring-Down Instrument
Karim Lakhani
Mentors: Geoffrey Smith, Al Fischer, Austen Scruggs, and Mitchio Okumura
Ambient aerosols are one of the largest contributors to air pollution, and determining their optical properties allows
scientists to create better models to study air pollution. However, because of their low extinctions (107 1 ),
special techniques besides UV-Vis spectroscopy are necessary to detect them. We attempted to build a three
wavelength (red, green, blue) cavity ring-down spectrometer to measure extinction by these particles. Cavity ringdown uses ultra-high reflectivity mirrors to create a large path length to detect molecules with small absorbance
cross sections. This project focused on coupling the three lasers into the same cavity and interweaving them to
obtain information for each wavelength at high repetition rates. I worked on setting up the spectrometer, writing
the Labview program that modulated the lasers and calculated the exponential fits in real time, and calibrating the
spectrometer to known gases. Future studies involve using the cavity ring-down spectrometer outside of the
laboratory setting to detect ambient aerosols.
Electrochromic Inverse Opals for Smart Windows With Static and Dynamic Optical Transmittance
Christian Lau
Mentors: Julia Greer and Victoria Chernow
Most smart windows employ electrochromic coatings with tunable optical transmittance upon the insertion and
extraction of charges. Incorporation of electrochromic coatings into windows have both reduced heat transfer into
buildings and enhanced their aesthetic appearance. Nevertheless, electrochromic smart windows can be further
improved by generating a photonic bandgap that continually reflects light in the near-Infrared range. Such a
photonic bandgap can be generated by an array of periodically altering refractive index. Periodic inverse opal
structures have been shown to produce complete photonic bandgaps in a wide range of frequencies. A number of
groups have reported scalable methods for fabricating arrays of titania inverse opals. My project aims to employ
these methods to produce the first smart window device with both dynamic electrochromism in the visible light
spectrum and a static photonic bandgap in the near infrared range. I investigate several sol-gel methods that

42

employ vertical evaporative deposition, drop-casting, and doctor-blade coating to optimize the fabrication of
inverse opals. Characterization of the inverse-opal morphology, chemical composition, and optical transmittance is
carried out to evaluate the performance of the device. Further study is needed to determine the most scalable
fabrication strategy to manufacture inverse opals for industrial production.
Ruthenium Catalysts With Chelating Fluoro-adamantyl-substituted N-Heterocyclic Carbene for
Z-selective Olefin Metathesis
SiiHong Lau
Mentors: Robert Grubbs and Brendan Quigley
Olefin metathesis emerged as a powerful method for the construction of carbon-carbon double bonds in a wide
variety of applications. However, achieving high Z-selectivity through olefin metathesis persisted as a challenge
due to the thermodynamically preferred E-olefins favored by most catalysts. A series of Z-selective metathesis
ruthenium catalysts with chelating fluoro-adamantyl groups has become our synthetic target in order to study how
the electronic nature affects the stability, stereoselectivity, and reactivity of the catalysts. The fluoro-adamantyl
amine has displayed a number of unexpected synthetic challenges, including interesting reactivity which we are
trying to characterize. This resulted in us employing multiple synthetic strategies for the preparation of the
catalysts.
Widefield Imaging of the Mouse Visual Cortex
Minh Nhat Le
Mentors: Doris Tsao and Tomo Sato
Calcium imaging using GCaMP6 mice is a promising tool to study how the mouse visual system represents surfaces
and objects. Before applying this technique to these higher-order visual processes, we first performed widefield
imaging to confirm that our newly assembled setup is working optimally. We performed thinned-skull surgery on
15 mice, and showed these mice a moving bar with a contrast-reversing checkerboard pattern to obtain a
retinotopic map of the early visual areas. Despite a few differences, the resulting retinotopic maps closely
resembled previously constructed maps from other studies, indicating the reliability of our measurements. Further
improvements to our existing setup, including the implementation of eye-tracking, will be carried out before the
technique can be used in future studies.
Predicting 3-Dimensional Structures and Binding Interactions of Human Olfactory Receptors
Morgan Lebby
Mentors: William A. Goddard, III, and Soo-Kyung Kim
G protein-coupled receptors (GPCRs) are a large family of trans-membrane proteins that mediate senses such as
odor taste, vision, and pain in mammals. Olfactory receptors (ORs) represent a large subset of GPCRs, with over
400 ORs identified in humans; however, the 3-dimensional structure and binding characteristics of most ORs have
not been determined. Through implementation of the GEnSeMBLE computational method, the most energetically
favorable conformations of ORs OR1A2-4 and OR2W1-5 were predicted, using Monte Carlo sampling of the , ,
and helical angles. These angles define the orientation of each of the seven trans-membrane helices that make
up all ORs. Out of the top 25 most energetically favorable structures, three were taken through the DarwinDock
method, which predicts the binding poses of the odorants that have been experimentally determined to cause
GPCR activation. From the docking results, we were able to determine the key residues that are involved in odorant
binding.
Expanding Fine-Grained Categorization to Novel Domains
Donsuk Lee
Mentors: Pietro Perona and Steve Branson
Many of the state-of-art techniques of fine-grained categorization depend upon part annotations for learning object
parts that help identifying subtle differences between subordinate categories. However, the high cost of acquiring
part annotations makes it challenging to expand fine-grained categorization to diverse domains. The goal of this
project is to develop a scalable fine-grained visual recognition system that does not require part annotations. The
system leverages the recent state-of-art technique, which generates parts using co-segmentation and alignment.
We demonstrate the effectiveness of this approach by evaluating on the Stanford cars-196 dataset. The complete
pipeline will be integrated into a web application where users can upload photos to retrieve relevant information.
We expect the system to be easily extendable to other domains with less cost.

43

Designing Immersive Visual Stimuli for Human Experiments

Ga Yeong Lee
Mentors: Ralph Adolphs and Juri Minxha
The typical approach a scientist would use to understand brain function is to record some kind of activity
(behavioral response, EEG, intracranial recordings, head tracking, eye tracking, etc.) while showing a stimulus in a
controlled and structured way. In a simple experiment, this may amount to showing a single image to a subject
while you record their brain activity using fMRI. This conventional approach however, rarely resembles the rich
nature of our experience of the world around us. In this project, we are trying to create a platform for addressing
scientific questions related to brain function, that is much more engaging in nature. We use a game engine to
develop scenes (i.e. stimuli), which can be experienced by the user in a virtual reality setting. We show, using a
simple exploration experiment, that users are much more engaged by this new stimulus presentation method and
that the data that is collected is much more abundant and complex than what is typically collected with on-screen
stimulus presentation.
Implementation of Dust Dynamics for Galaxy Formation Simulations
Hyunseok Lee
Mentor: Philip F. Hopkins
We study the behavior of large dust grains in turbulent molecular cloud. Interstellar dust plays a crucial role in
many important processes in galaxy evolution, including conversion from neutral to molecular hydrogen and
formation of low-mass stars in the early universe. While interstellar dust grains are often modeled to move with
neighboring gas flow in numerical simulations, the dust grains in typical molecular cloud does not necessarily move
with the gas. We therefore directly simulate dust dynamics in highly supersonic turbulent box. Under these
conditions, we study how grain size and flow velocity affect local densities of dust and gas. The results exhibit
dramatic fluctuations in the local dust-to-gas ratio, thereby implying large small-scale variations in abundances,
dust cooling rates, and dynamics. Further researches can incorporate Lorentz forces and other physics neglected
here.
Characterization of Tumor-Infiltrating T-cells With Single-Cell Barcode Chip Technology in the Context
of Melanoma Therapy
Jihoon Lee
Mentors: James Heath and Jing Zhou
Melanoma is the most lethal of skin cancers. There are a multitude of therapies being developed to combat it,
among which is the amplification of a melanoma patients natural immune response tumor-infiltrating
lymphocytes (TILs). TILs have the ability to attack and help destroy tumor cells due to their distinct active
signaling pathways. However, often the natural response is too small for there to be effective elimination and in
fact may induce tumor cell de-differentiation in order to resist the attacks. In order to detect and amplify such a
population, TILs cell surface receptor populations may be quantified. In order to accomplish this for individual cells
and many receptors efficiently, the single-cell barcode chip has been developed to assay single lymphocytes in a
multiplexed manner. This investigation examines new uses of SCBC technology to characterize different aspects of
tumor-infiltrating T-cells, entailing the design and construction of SCBCs in combination with a model of tumorinfiltrating T-cell clustering and signaling behavior and a SCBC analysis of melanoma cell proteomic responses to
various drug and TIL therapies, with the ultimate goal of developing a more effective melanoma therapy.
Immersive and Collaborative 3D Scientific Data Visualization
Joon Lee
Mentors: George Djorgovski and Ciro Donalek
Scientific data sets are growing in both volume and complexity, providing novel opportunities for discovery. These
massive data sets, however, are not fully exploited without tools we can use to discern the hidden patterns in the
data. As a result, data visualization is becoming an increasingly indispensable part of scientific research. The
Center for Data-Driven Discovery is developing a visualization program, iViz, that allows users to immerse
themselves in a virtual reality containing data points which can represent numerous dimensions using position,
shape, size, color, and texture. This project seeks to improve iViz through the addition of several new features.
First, a spherical plot was added to make the program suitable for geographical and celestial data. Additional
changes were made to make visualization and pattern recognition effortless in these spherical plots. In addition,
several types of grids and planes were created to aid in data analysis. Finally, a networking framework was
developed to allow for collaborative visualization through the Web.

44

A Partition Function Algorithm for Nucleic Acid Secondary Structure With Dangle and Coaxial Stacking
Yoon Lee
Mentors: Niles Pierce and Nicholas Porubsky
We describe an algorithm for performing thermodynamic analyses of interacting nucleic acid strands. Given a
nucleic acid sequence, we can calculate the partition function, which describes the statistical properties of a system
in thermodynamic equilibrium, over secondary structure space with the dynamic programming paradigm. We can
determine the partition function of a given sequence over all unpseudoknotted secondary structures in O(N4) time,
which can be reduced to O(N3) complexity, where N is the length of the strand, if we simplify the loop-based free
energy model. The objective of this study is to develop a partition function algorithm that rigorously incorporates
the energy contributions of dangle and coaxial stacking states as they provide additional energetic stabilization for
multibranch and exterior loops. We update the recursions for partition function calculation in NUPACK, a software
suite for the analysis and design of nucleic acid systems, to quantify how these two factors are relevant in
improving accuracy in conducting a thermodynamic analysis of a dilute solution of interacting nucleic acid strands.
We use the internal validation and explicit enumeration schemes to ensure the correctness of the algorithm.
A Wavelet-Based Finite Element Approach to Atomic-Scale Mechanics
Daniel V. Leibovici
Mentor: Michael Ortiz
Numerical simulations of microscopic material are often performed at an atomistic scale in order to study
phenomena like crack-tip or dislocations. However due to their high computational cost, multi scaling and mesh
reduction methods have become key points in computational mechanics. This research project deals with the
implementation of a new method to perform mesh coarsening, based on the wavelet theory. Thanks to the
properties of hierarchical refinement of the scaling functions, a lattice can be reduced or refined hierarchically on
different levels of precision. The aim is to study how a phenomenon, like a wave, would propagate in uniformly fine
and coarse models, and in the future in a combination of both. The study of wave-reflections between areas of
different level of refinement, and the development of new integration methods to avoid them opens a broad area of
research: the aim is the development of a revolutionary method of simulating molecular dynamics which would
highly reduce the current computational costs.
A New High-Frequency Search for Galactic Center Millisecond Pulsars Using DSS-43
Cameron R. Lemley
Mentors: Thomas A. Prince and Walid A. Majid
The primary 70-meter Deep Space Network antenna (DSS-43) in Canberra, Australia was equipped with a new
high-frequency (18-28 GHz) receiver system in May 2015 for use in a search for Galactic Center (GC) millisecond
pulsars. The primary motivation for this search is that a pulsar in the Galactic Center region (especially one that is
gravitationally bound to the massive black hole at the GC) would provide unprecedented tests of gravity in the
strong-field regime and would offer an entirely new tool for probing the characteristics of the Galactic Center
region. Preparation for the GC pulsar search has involved the development of a single-pulse search pipeline that
integrates tools from both Fortran and Python as well as the implementation of this pipeline on high performance
CPUs. The original version of the search pipeline was developed using Vela Pulsar data from DSS-43, and a more
refined version that relies upon chi-squared fitting techniques was ultimately developed using Crab Pulsar data.
Future work will involve continued testing of the single-pulse search pipeline using data from the rotating radio
transient (RRAT) J1819-1458 as well as the characterization of pulses from this system using high time resolution
data from the new receiver system on DSS-43.
Effect of Exogenous PU.1 and PU.1-ets on Commitment and Spi1 Autoregulation in Hematopoietic
Progenitor Cells
Jessica D. Li
Mentors: Ellen V. Rothenberg, Hao Yuan Kueh, and Jonas Ungerbck
By the double negative (DN)2b stage of T cell development, hematopoietic progenitor cells can no longer
differentiate into other cell types. Here we study PU.1, an Ets family transcription factor encoded by Spi1 that is
involved in commitment. PU.1 positively regulates Spi1, so cells with high PU.1 should maintain similar expression.
However, while necessary to initiate development, PU.1 levels drop sharply during the DN2b stage. To study PU.1
autoregulation change, we infected precursor cells with H2B (control), Spi1, or PU.1-ets, the PU.1 DNA binding
domain, and grew them in either T cell supporting (OP9-DL1) or non-supporting (OP9-Mig) environments. We
found that cells infected with PU.1 on OP9-DL1 had similar levels of endogenous PU.1-GFP and T cell development
as control cells. On OP9-Mig, exogenous PU.1 impaired B cell development in favor of CD11b+ cells. Adding PU.1ets to cells biased them toward CD11b+ on both environments. In samples with higher proportions of CD11b+ cells,
CD11b+ populations with low PU.1-GFP levels not found in control samples appeared. Together, these results both

45

support the idea that PU.1 no longer positively autoregulates itself in T cells suggests a more complex role of PU.1,
as the simple act of PU.1-ets binding to cis-regulatory elements appears to influence fate decisions. Further work
will characterize the currently unknown CD11b+ PU.1-GFPlow+ populations and determine which genes are affected
in these PU.1-ets infected cells.
Automatic Data Collection for Real-Time Usability Problem Detection in CRISIS Virtual Environments
John Li
Mentor: Ebba Hvannberg
Crisis training is necessary to prepare medical first responders for emergency situations. In order to train for these
situations, an environment is often prepared with the appropriate props and actors. However, it takes a lot of
careful planning, time, and money to set up these scenarios for the frist responders. The CRISIS project addresses
this by creating virtual environments for first responders to train with, saving the time and money used to create a
real-life scenario. However, virtual environments must also closely replicate real-life environments if virtual training
is to be effective. Therefore, my task is to automatically detect usability problems in these virtual environments
during a training scenario. In order to detect these problems, data from the environment must be undergo
Complex Event Processing (CEP), which is done through the software product known as Esper and written Event
Processing Language (EPL) scripts. The data from the environment must also be sent to Esper, which is done
through the use of a message broker known as Apache ActiveMQ. With this architecture, we expect that our virtual
reality environment will be able to detect usability problems in real-time training.
An Automated System for Geometric Differentiation of Mesh-Based Energies
Kevin Li
Mentors: Eitan Grinspun, Keenan Crane, and Alan Barr
We propose an automatic geometric differentiation (AGD) system that produces meaningful geometric expressions
for derivatives of mesh-based quantities. Numerical evaluation of forces and gradients is a ubiquitous problem in
physical simulation and geometric optimization, yet expressions quickly become cumbersome for energies involving
geometric quantities associated with surface or volume meshes. Obtaining simple, meaningful expressions typically
requires high-level geometric intuition that current computer algebra systems lack. Yet consistent patterns in byhand derivations suggest that an intelligent AGD system may be within reach. For instance, one often infers the
gradient direction by identifying the directions along which the energy remains constant, a procedure that can be
performed by brute-force computation. Currently, we extend established automatic and symbolic differentiation
methods to incorporate mesh-based geometric quantities as fundamental nodes in an expression tree. This
extension prunes the tree, and facilitates the use of expert geometric knowledge during simplification as well as
intelligent caching. Preliminary experiments indicate that even basic modifications to nave automatic
differentiation can yield over a 2x reduction in computational overhead; gains from modifying symbolic
differentiation are expected to be strictly better. Ultimately, this tool aids rapid development of fast, reliable
numerical code, allowing researchers to focus more on high-level exploration.
Audiovisual Illusory Rabbit: Perception Across Senses and Time
Monica Li
Mentors: Shinsuke Shimojo and Noelle Stiles
Unlike the traditional five senses, we do not have a specific sensory organ to sense time; time seems to pass fast
or slow, and memories can be difficult to keep in chronological order. Furthermore, while the timing of each sense
appears to be distinct, these separate sensory modalities are fused, giving rise to illusions such as the multimodal
rabbit.
The audiovisual rabbit occurs when two flashes and three beeps are spaced apart spatially and temporally, with the
first and third beep paired with a flash to the left then one to the right, respectivelyeach of the flashes paired with
beeps flanking (in space and time) a single flash, with the standalone second beep temporally between them. The
brain stitches together the impression that the two stimuli modalities are actually one in continuous movement,
and thereby generatinges an illusory flash with the unpaired beep.
This rabbit is particularly interesting because the illusory flash is postdictive;, in that the brain only knows the
beep-flash pairing pattern after the illusory flash is supposedly perceived. We have reproduced the audiovisual
rabbit illusion and begun exploring changes in color and shape.
Additionally, we have found a new postdictive suppression illusion, where participants perceive three flashes with
two beeps paired with the first and third flash as only two flashes.

46

Optimising SUSY Search at CMS With Machine Learning Techniques

Yuting Li
Mentors: Maria Spiropulu and Javier Duarte
The Higgs detection with its 126 GeV mass is hinting at physics beyond the standard model. We are at the stage
where we are unclear as to what discovery story awaits and need to cater for an extremely small amount of
signals. There have been studies demonstrating the use of supervised machine learning techniques in signal
detection but supervised methods have the risks of overtraining on Monte Carlo imperfections and the difficulties of
getting statistical interpretations. In this paper, the Self Organising Map (SOM) was applied as an unsupervised
data-driven clustering algorithm. To tackle the problem of the lack of statistics, the Neural Autoregressive Density
Estimator (NADE) was first introduced for background estimation in the cells in the SOM. Its ability to detect
outliers was also investigated in this paper.
Examining the Effect of Deformation Radius on Zonal Jets Based on a Quasi-Geostrophic Model
Ziwei Li
Mentors: Andrew P. Ingersoll and Cheng Li
In the study of planetary atmospheres, 2D turbulence model with parameterized moist convection plays an
important role in investigating the interaction between moist convection and large-scale circulation. In an earlier
study, members of this group (Liming et al., 2005), used a quasi-geostrophic model to study the dynamics of
Jupiter and Saturns atmospheres. Their results differ from others (Scott et al., 2007, Showman et al., 2006) in
that the deformation radius had little impact on the width and amplitude of zonal jets. After a preliminary study, we
find several unphysical treatments in the original numerical formulation that might lead to the above disagreement,
including the global adjustment of the streamfunction and the boundary layer forcing method. We changed those
treatments and our new model produces more reasonable results. We show that the deformation radius may
indeed affect the zonal velocity profile, and at the same time, there is possible northward migration of the zonal
jets, which awaits further investigation. Later on, we will keep investigating the possible effects of moist convection
parameterization, spatial resolution dependence, and baroclinic stability in our newly formulated model.
The Carbery Conjecture and Related Problems
David J Lichko
Mentor: Majid Hadian-Jazi
Gauss's quadratic reciprocity law, which is a fundamental result in algebraic number theory, explains how quadratic
polynomials decompose modulo primes. It says that the decomposition depends only on congruence conditions on
the prime. In general, no such law can describe the decomposition of a cubic polynomial modulo primes. We
present reciprocity laws for certain families of cubic polynomials. When the discriminant D of the polynomial is
fixed, we formulate a reciprocity law in terms of congruence conditions on solutions to a finite number of quadratic
forms. If the class number of Q(-3, D) is not divisible by 3, the law can be explicitly written in terms of the
coefficients of the cubic polynomial. When the prime modulus is fixed instead, we show how the decomposition can
be determined by a congruence condition on the coefficients of the polynomial. We also study the case where the
splitting field of the cubic is in the Hilbert class field of Q(D).
Fabrication of Microlens Arrays Through the MicroAngelo Technique Using Thermocapillary Forces
Soon Wei Daniel Lim
Mentors: Sandra M. Troian and Kevin R. Fiedler
Plano-convex microlens arrays were fabricated using the one-step MicroAngelo technique, which uses strong
thermal fields to modulate forces driven by surface tension gradients on a molten polymer nanofilm interface to
achieve the desired surface topology. Spatially inhomogeneous temperature gradients were achieved by varying
the heights of photoresist elements on a cold plate placed above the polymer film. The fabricated arrays will
implement a Shack-Hartmann wavefront sensor, and its optical quality will be compared to that of a commercially
available microlens array. The experimental results thus far show that with suitable control over the thermal field,
MicroAngelo has the potential to form a diverse spectrum of ultra-smooth surface topologies in a single non-contact
fabrication step.
Measuring Micro-Mechanics of the Drosophila melanogaster Wing Hinge With in vivo Fluorescent
Muscle Imaging
Amanda Lin
Mentors: Michael Dickinson and Theodore Lindsay
Although we are gaining a greater understanding of how flies use sensory signals to execute flight behaviors, the
exact mechanical means controlling the wing motion remains a mystery. Previous studies have measured activity
and strains of a few muscles controlling the wing-hinge; however, these studies performed under a restricted set of
sensory conditions. To expand on these studies, we took advantage of transgenic Drosophila melanogaster that
express fluorescent proteins and measured the strains in several identified muscles of the wing hinge. We

47

combined fluorescence microscopy with high-speed video and stroboscopic measurements of the steering muscles
during both quiescence and visually-elicited flight maneuvers. As such, we were able to detect visually evoked
changes in the strain cycles of the tergo-pleural muscles and first basalar muscle. From our analysis, we have
reason to believe that the first basalar muscle is correlated with wing-stroke amplitude as, in many cases, the two
occur with the same frequency. We also see that the strain of the tergo-pleural muscles are proportional to the
wing-beat frequency and are important during the onset of flight. Thus, with these experimental methods, we can
continue to work to understand how the steering muscles work together to control wing motion.
Analysis of K2 Light Curves and Spectral Energy Distributions to Detect Earth-Like Exoplanets
Hanzhi Lin
Mentors: Rachel Akeson and Erik Petigura
Since the K2 mission launched, there has been over 100,000 stars discovered, ranging from M dwarfs to red
giants. Light curves of the stars display the energy flux over time, thereby allowing astronomers to identify
exoplanet transits as parabolic dips. Unfortunately, different types of astronomical and arbitrary noise are prevalent
in these light curves. Using Python, certain filtering functions were coded to remove the noise, thereby allowing
one to easily identify transits. After pinpointing certain planet transits, the period, size, and composition of the
planets could be estimated. The spectral energy distributions (SED) of the stars were also computationally
analyzed to test the accuracy of certain photometric catalogs. By comparing the stars SED plots from all of the
observational catalogs to those from reliable theoretical models, particular catalogs were revealed as flawed. After
removing suspect catalogs, the stars temperatures, luminosity, and other physical properties were computed. With
the observation of planet transits in light curves and computational calculations of the stars properties, exoplanets
with Earth-like attributes could potentially be discovered.
Adding a Treatment for CO2 to pMELTS
Zachary Lipel
Mentor: Paul Asimow
pMELTS is a software that models magmatic equilibria at high pressures. It builds upon its predecessor, MELTS,
which functions similarly up to about 1 GPa, whereupon it is inaccurate. Recently, an H2O-CO2 mixed fluid
saturation model compatible with rhyolite-MELTS (an updated version of MELTS) was developed. We aim to apply a
similar method to add a treatment for carbon dioxide to pMELTS. I organized and uploaded a collection of data
containing carbonate (both liquid and solid) phases. The data define the stability fields of carbonate,
multicomponent liquid carbonatite liquids, and coexisting silicate and carbonatite liquids in pressure-temperaturecomposition space. Using existing thermodynamic and solution models for silicate liquids at high pressure and for
carbon dioxide and calcium carbonate in those liquids at low pressures, the aforementioned data should allow for
the calibration of a thermodynamic model that correctly accounts for the behavior of carbon dioxide saturated
melts at high pressures. We will evaluate the model by both examining energy residuals and by using forward
modeling of carbonate bearing experiments to check the temperature of phase boundaries, the progression of
phase proportions, and predicted phase compositions.
Semisynthesis and Analysis of AKT1 Phosphoforms
Albert Liu
Mentors: Philip Cole, Nam Chu, and Pamela Bjorkman
Akt1 is a protein kinase that plays an essential role in cell signaling, regulating multiple cellular processes such as
apoptosis and cell proliferation. Notably, dysregulation of Akt1 signaling has been shown to lead to cancer. It has
been well established that the activity of Akt1 is primarily regulated by phosphorylation at two residues: T308 and
S473. Recently, two new phosphorylation sites at the C-terminus of Akt1, S477 and T479, were discovered.
However, the function of these two new sites of phosphorylation has not been characterized to a large extent. By
taking advantage of a natural cysteine residue within Akt1 (C460), we were able to use native chemical ligation to
semisynthesize four different phosphoforms of Akt1: no phosphorylation, pS473, pS477 + pT479, and pS473 +
pS477 + pT479. We are currently assaying the different phosphoforms with regard to two different activities:
ability to be activated by PDK1 (the enzyme which installs pT308), and ability to activate Akt1 substrates. By
comparing the results from each of the different Akt1 phosphoforms, we hope to unravel the function of the newly
discovered phosphorylation modifications. Lastly, we also hope to discover novel substrates of fully phosphorylated
Akt1 by utilizing a protein microarray strategy.
Timing Resolution Studies of Hamamatsu Silicon Photomultipliers
Eric Liu
Mentors: Maria Spiropulu, Si Xie, Artur Apresyan, Anatoly Ronzhin, and Cristian Pea
Photodetection technology has wide application in modern scientific and commercial fields, from spectrometry to
subatomic particle detection. Timing precision of detectors is critical in the performance of modern particle
detectors, which calculate particle masses using strong magnetic fields and precise time-of-flight measurements.
Silicon-based photodetectors are currently in use in the Large Hadron Collider (LHC), since silicon's performance

48

and timing precision are largely unaffected by strong magnetic fields. The current detectors have a time resolution
of 200-300 ps, which is an order of magnitude too large to handle the increased collision rate of the next LHC
upgrade. Sample detectors from the next generation of silicon photodetectors were tested using picosecondduration laser pulses.
Engineering Bispecific Antibody-Like Reagents Against HIV
Phillip Liu
Mentors: Pamela J. Bjorkman and Rachel P. Galimidi
Despite more than three decades of effort, no effective vaccine for HIV has been developed, and AIDS remains a
serious global public health issue. HIV presents numerous difficulties to researchers with its high mutation rate,
glycan shielding of potential epitopes, as well as shielding constant regions of glycoproteins with highly variable
regions to further protect against antibody binding. Additionally, the extremely low density of HIV envelope spikes,
around 14 for every virion, results in few opportunities for inter-spike linking IgG bivalent binding, and intra-spike
bivalent binding is not possible due to the structure of IgGs. Recent research has resulted in the development of
protein-DNA structures where dsDNA links two Fc antibody regions. These reagents exhibit intra-spike bispecific
binding. In this project, we design and engineer bispecific all-protein structures with protein linkers in place of DNA
via controlled Fab-arm exchange. HIV pseudovirus neutralization assays were performed to assess the potency and
breadth of the constructs. We hope to perform polyreactivity ELISA as well as additional neutralization assays to
further analyze these reagents.
Integration of Behavioral Commands for Improved State Estimation and Control of Pioneer 3-DX for a
Resilient Spacecraft
Sandra Liu
Mentors: Richard M. Murray and Catharine L. McGhan
In the upcoming years, scientists and researchers want to explore new areas in space exploration: places that are
more distant, enigmatic and thus less accessible. This requires stronger and more resilient spacecraft that will be
able to be risk-aware and act autonomously when commands are unable to be provided. To deal with these
challenges, a Resilient Spacecraft Executive (RSE) architecture is being designed so that space robots can adapt to
equipment failures, deal with different environments and make such risk-aware decisions. This project deals with
developing and implementing algorithms to support the underpinnings of the RSE and vehicle objectives like
obstacle avoidance and other higher-level processes. Enabling the robot to respond to autonomous commands,
implementing motion constraints on vehicular movement and putting together an Inertial Measurement Suite (IMU)
sensor suite that can supply more useful data will help to support future endeavors to make space vehicles more
autonomous and intelligent.
Rapid Estimation of Earthquake Magnitude and Epicenter Using Ground Motion Parameters
Timothy Liu
Mentors: Egill Hauksson and Jennifer Andrews
Earthquake early warning systems (EEWS) are designed to provide several seconds to tens of seconds of warning
that shaking from an earthquake is imminent at the user location. Such systems use a network of seismometers
scattered across an area with active faults to constantly monitor and report ground motion. When an earthquake
begins, algorithms must rapidly analyze ground motion data and estimate the epicenter and magnitude of the
event. If an accurate estimate can be made quickly enough, an alert can be issued, possibly giving individuals
crucial seconds to find shelter. EEWS are dependent on algorithms that can quickly and accurately make magnitude
and epicenter estimates. One of the algorithms used by ShakeAlert, an EEWS being developed for the West Coast
of the United States, is the TauC-Pd On-Site algorithm. On-Site currently uses the ground motion parameter TauC
to estimate the magnitude of an event using a single station. This paper proposes an alternative method for
estimating the magnitude of an event at a single station that uses the attenuation of peak displacement and peak
velocity with distance from the epicenter. Results suggest that this attenuation model produces more accurate
magnitude predictions than those made by using TauC. Additionally, a new method for more accurately
determining the epicenter of an event based on p-wave arrival times is also proposed. Analyses of test data show
that the two methods provide improved performance and robustness of the OnSite algorithm.
Investigation of the Regulation of SRP-Mediated Co-Translational Protein Targeting
Yun (Demi) Liu
Mentors: Shu-ou Shan and Jae Ho Lee
The signal recognition particle (SRP) provides strict coordination between protein synthesis and translocation, and
the interaction between SRP and its receptor (SR) is required for the initial targeting of the ribosome-nascent chain
complex (RNC) to the ER to form the ribosome/translocon junction, thereby ensuring proper biogenesis of proteins
in the crowded cellular environments. We are interested in accurately describing the kinetics and thermodynamics
of the interaction between SRP and SR, and testing if and how the Alu-domain communicates with the S-domain of
SRP during recognition of the RNC and its targeting to SR. To examine these, we will establish Forster Resonance

49

Energy Transfer (FRET) assays that enable to monitor individual molecular interactions in the SRP pathway at high
resolution. SR protein has been expressed and purified; human SRP54 has been labeled with a Cy3b fluorescence
probe. Upon successful fluorescence labeling of SR, monitoring FRET between SRP and SR will provide insights into
their interaction. Using the FRET pair of SRP9 and the RNC, the energetics of the SRP-RNC binding will be
investigated to determine the mechanistic characteristics of the Alu-domain in SRP-mediated protein targeting
pathway.
Random Approach For Distributed Reed-Solomon Code Construction Using Balanced Column Sparsest
MDS Code
Zihan Liu
Mentor: Babak Hassibi
This project looks at the construction of distributed Reed-Solomon code, expecting this Reed-Solomon code's
subcode to be full rank and to obtain the upper bound of at the meantime. Especially, we concentrate on the
balanced column sparsest MDS code, and anticipate to randomly assign the positions of zeros in each row, or
equivalently, the permutations of field elements of Reed-Solomon code columns. By revealing the generator matrix
to both ends of the channel after this random construction, we could decode it efficiently.
Mechanoluminescent Inverse Opals: Design and Fabrication
Peter Lommen
Mentors: Julia Greer and Victoria Chernow
Photonic crystals, by virtue of their nanoscaled periodic architecture, act as natural optical bandgaps, selectively
amplifying certain wavelengths of light. Photonic crystals occur naturally in nature, often in the form of opals, made
of nano-sized spheres of silica (SiO2) which self-assemble into a hexagonal close-packed lattice structure. The
structural inverse of an opal can be created by infiltrating the gaps with a polymer mix, and then etching out the
silica spheres of the original opal. These inverses, so-called inverse opals are studied by multiple projects
undertaken by the Greer group. This project's focus is the design and fabrication of a mechanoluminescent
photonic crystal, which is an elastomer-based inverse opal, functionalized with nanocrystals of metal-doped zinc
sulfide, a mechanoluminescent material. When subjected to mechanical stress, the zinc sulfide particles emit light
of a specific wavelength, which can then be amplified by the inverse opal's structure. After successful fabrication,
the inverse opal will be optically characterized and fine-tuned for optimal performance.
Life Cycle Analysis of Electric Vehicle Systems
Mark Lorden
Mentor: Guillaume Blanquart
One of the most important considerations to take when planning a sustainable vehicle system is the environmental
impact of the vehicle. Life cycle analysis is a method of quantifying the environmental impact of an engineering
system, taking into account the power efficiency of the system, as well as the environmental impact of the
materials used to construct the system. This analysis can then be used to find ways of lessening the environmental
impact of the system.
In Vitro Glucose Analysis Utilizing Raman and Millimeter Wave Spectroscopy
Nathan Louie
Mentors: Andrei Faraon and Peter Siegel
Accurate, non-invasive methods for continuous monitoring of blood glucose levels has been a long-term goal for
many healthcare researchers and companies. The current gold standard for measuring glucose involves direct
sampling of the blood. This means continuous readings all over the body is not possible. By coupling the theory of
transmission loss in millimeter waves as a determinate for glucose concentration and optical Raman Spectroscopy
to analyze the spectral signal of the solution it is possible to determine the concentration of glucose in a solution
for a period of time. The millimeter wave portion is set up with a network analyzer and waveguides to send an
initial frequency through the sample. This will result in a lower frequency return. The Raman optical set up will be
using an excitation source, a laser, and reading the resulting spectral signal that is reflected off the sample. The
Raman spectrum of the scattered excitation coupled with the transmission loss will then be analyzed to determine
and more specific result considering the signal from both techniques are fairly weak. As of now, we have only
obtained results for the millimeter wave portion and, like predicted, the signal is almost the same as the original
frequency. However, we do plan on moving forward to testing on rats after we complete the Raman spectroscopy
of the glucose samples.

50

Characterization of Heavy Crude Oil Fractions by 13C-NMR Spectroscopy

Alison Lui
Mentors: William Green, William Goddard, Soumya Gudiyella, and Lawrence Lai
Supercritical water (SCW) has recently been found to be an effective catalyst for the desulfurization and cracking of
crude oil with less coke formation than current dry cracking methods. To better understand the role of SCW in this
complex decomposition, we seek to quantify the rates of production and destruction of specific functional groups in
crude oil and in the SCW reaction products. Here, we present carbon-13 nuclear magnetic resonance spectroscopy
(13C-NMR) as a promising method of heavy crude analysis. Unlike separation by gas or liquid chromatography,
13
C-NMR does not have restrictions on molecular weight or polarity and also provides information about the
samples in terms of their functional groups. In this project, we investigated the feasibility of 13C-NMR as a method
of characterization for crude oil and determined the optimal parameters of spectra acquisition. In our heavy crude
spectra, distinct aliphatic functional group peaks could be characterized; however, peaks in the aromatic and
alkene regions were visible but broad and ill-defined. We remain optimistic that better definition can be achieved
by increasing the number of scans taken for a single spectra and by modifying the processing parameters and
conclude that 13C-NMR still has promise as a crude oil characterization tool.
Design and Fabrication of Durable Nano Vacuum Triodes With Low Turn-On Voltages
Daniil Lukin
Mentors: Axel Scherer and William M. Jones
Four-terminal planar vacuum triodes with emitter-collector gaps under 10 nm were fabricated out of tungsten on
sapphire. Devices display numerous desirable characteristics: maximum sustainable current of 0.5A, modulation
at 5V, gate and emitter leakage at base noise level, and no visible degradation after an hour of operation. Yield
rate of 20% was obtained with our fabrication method. Vacuum triodes are expected to operate at high
temperatures and high frequencies, the latter being particularly desirable for integration with on-chip photonics.
An Analysis of the Novel Cytoplasmic Streaming Patterns of Caulerpa
Catherine Ma
Mentors: Elliot Meyerowitz and William Gibson
It has long been accepted that unicellular organisms lack the complexity and evolutionary advantages that
multicellularity entails. Caulerpa is genus of seaweeds that seems to defy this generalization. Even though
members of the genus are unicellular, they possess metabolic complexity on par with that of multicellular
organisms, can grow to be quite massive, and thrive in many underwater environments. Studies have shown that
the cytoplasmic streaming patterns of Caulerpa are relatively distinct from those of typical multicellular plants, and
we speculate that studying them can give us insight into how Caulerpa can prosper albeit its unicellularity. To
determine its unique cytoplasmic contents, we cultivated Caulerpa prolifera, extracted its DNA and mRNA, and
implemented many bioinformatics algorithms on the data to assemble the genome and transcriptome of the
species and discern its gene expression. We also took images of C. prolifera leaf segments and utilized many image
processing techniques to quantify the cytoplasmic flow patterns in the leaves of the plant. We ultimately hope to
uncover cytoplasmic elements that contribute to Caulerpas unprecedented success in the wild, and subsequent
studies of the plant will focus on how these elements can provide unicellular organisms with the survival
advantages of multicellular organisms.
On-Demand Power Source for Medical Electronic Implants: Harvesting Energy From AcoustoMechanical Vibrations of Human Vocal Cords
Yuan Ma
Mentor: Hyuck Choo
Microelectromechanical systems (MEMS) and microelectronics have presented a wide variety of medical
applications due to the convenience of their small size and low power consumption. However, since these
implantable electronic devices must be continuously running in the background, their batteries ultimately need to
be replaced or recharged, a process that leads to additional invasive surgery and discomfort to patients. Therefore
we will investigate and create a device that can harvest energy from the acousto-mechanical vibrations of human
vocal cords. After characterizing the range of human voice frequency, we are designing and building a piezoelectric
energy harvesting setup that converts the mechanical energy of vibrations from human vocal cords at resonance
frequency into electrical energy. Using a system of multiple piezoelectric cantilevers placed in contact with the skin,
we can generate a small amount of power from vibrations resonating throughout the skull. Our energy harvesting
setup will serve as a power source to supplement or replace the batteries in implantable microelectronic sensors.

51

Kepler Target Analysis Through Light Curve Examination and Spectral Energy Density Modeling
Mason MacDougall
Mentors: Rachel Akeson and David Ciardi
Each K2 campaign carried out by the Kepler Space Telescope has continued to add to the list of exoplanet
candidates found through transit detection amidst stellar light curves. The majority of such planets have been
discovered using programs requiring a minimum of 3 events to confirm a periodic transit. Here, we discuss our
efforts to visually identify remaining single- and double-transit events within K2 campaigns 1-3. Our goal is to
analyze the light curves of the brightest 500 M-dwarfs and all stars with kepmag <10 from each of the specified
campaigns in search of long-period planetary transits. We have constructed several filtering tools which have
facilitated our detection of 12 candidate eclipsing binary star systems thus far. We have likewise performed stellar
spectral energy density modeling for the ~100,000 K2 targets through campaign 5. Such models are used to
obtain stellar parameters including surface gravity and effective temperature. Our aim for this portion of the
project is to verify the accuracy of our modeling method and construct a database for our models along with stellar
parameters established from these models. This database will offer a convenient location for finding the necessary
stellar parameters to conduct follow-up research on potential planets.
Structural Studies of HIV-1 BG505 SOSIP.664 Trimer/8ANC195 Fab/Soluble CD4 Complexes
Sumana Mahata
Mentors: Pamela J. Bjorkman and Haoqing Wang
Human immunodeficiency virus type 1 (HIV-1) is a lentivirus that threatens the public health of millions worldwide.
HIV-1 has been extensively studied, and its high mutation rates and glycan shielding prevents most individuals
from exhibiting any neutralizing antibodies against it. Recently, a few papers have described a soluble, stable and
cleaved HIV-1 gp140 trimer, known as BG505 SOSIP.664 gp140, whose structure closely resembles the native
state of the HIV-1 envelope glycoprotein. 8ANC195 is a broadly neutralizing antibody (bNAb) whose target includes
both the gp120 and the gp41 subunits of the HIV-1 envelope spike. We seek to investigate the structure of
complexes between the antigen-binding fragment (Fab) of 8ANC195 and BG505 SOSIP.664 trimer in different
conformations. We will generate and purify each component, form and purify complexes, and then collect single
particle data on the complex via negative-stain single particle electron microscopy. We will then process the
micrographs collected in order to determine the structure of the BG505 SOSIP.664/8ANC195/soluble CD4 complex.
We hope to confirm that the neutralizing mechanism of 8ANC195 involves keeping the trimer in the closed
conformation, thus preventing it from undergoing conformational changes required for membrane fusion, and to
prove that 8ANC195 is able to neutralize the trimer in different conformations.
Predictors for Persistent Atrial Fibrillation Over Paroxysmal Atrial Fibrillation: A Cross-Sectional
Analysis of Health eHeart
Chaitanya L. Malladi
Mentors: Gregory M. Marcus and Justin Bois
Numerous research studies have analyzed risk factors for atrial fibrillationa serious heart condition that affects
about one percent of the population in the United States. There are two main types of AF: persistent and
paroxysmal. Quality of life for patients has been shown to be worse in those with persistent AF versus those with
paroxysmal AF. Additionally, those with persistent AF are more frequently treated with procedures such as
cardioversions and potentially toxic anti-arrhythmic drugs. Understanding risk factors for each type of AF may help
health care providers better advise their patients on lifestyle modifications that can mitigate each risk.
Through analyzing the atrial fibrillation patient pool in UCSFs Health eHeart study, we examined age, sex, race,
hypertension, CAD, CHF, diabetes, sleep, family history, and alcohol consumption as predictors of type of AF. In a
multivariable model adjusting for age, sex, race, CHF, and diabetes, we found that increasing age (OR of 1.25 per
decade increase in age) is a predictor. Additionally, being male as opposed to female has an OR of 1.47. People of
the Asian or Pacific Islander race have a lower risk of acquiring persistent AF (OR of 0.297) compared to whites.
Lastly, congestive heart failure (CHF) is also a statistically significant predictor (OR of 1.76) of persistent AF. This
result accords with previous research indicating how structural damage to the heart strongly correlates with
persistent AF. Our analyses on sleep quality, alcohol consumption, hypertension, CAD, diabetes, and family history
yielded no statistically significant correlations for predicting type of AF.
Fourier Transform Spectrometer System Upgrade and Complex Impedance Studies of a
Semiconductor Bolometer
Aashrita Mangu
Mentors: Suzanne Staggs and Jamie Bock
In an effort to understand the origin of the universe by accurately imaging cosmic microwave background (CMB)
radiation emitted right after the Big Bang, the Princeton experimental CMB group has been working on numerous
experiments, including the current Atacama B-mode Search (ABS) and Atacama Cosmology Telescope (ACT), as
well as the Multimoded Survey Experiment (MuSE) using the detector developed for the proposed Primordial

52

Inflation Explorer (PIXIE) satellite mission at NASA Goddard Space Flight Center (GSFC). Both multimoded optics
and the Fourier Transform Spectrometer (FTS) have provided increased sensitivity for CMB polarimetry, allowing
better detection and characterization of the polarization signal from an early, inflating Universe. This two part study
is the result of the coupled necessity of multimoded optics and the FTS: the objective of the first part is to develop
a new user interface (UI) for the FTS to determine bandpasses of required detectors and components by
incorporating a producer-consumer state-machine software structure; the second part is to model the frequencydependent complex impedance and voltage responsivity of multimoded bolometers. While the updated UI will aid in
more efficient data collection and control of the FTS for the user, the latter is another step in the optical testing
required to further advance the MuSE and PIXIE projects and develop a better understanding of bolometer theory.
Mechanics of Peeling of Heterogeneous Medium
Om Vijay Margaj
Mentors: Guruswami Ravichandran and Owen Kingstedt
The measurement of interface properties between an adhesive layer and substrate can be studied through the peel
test. In this study thin extensible strips of PDMS are peeled from rigid substrate. The substrate consists of
patterned transparencies fixed to a rigid base. Heterogeneities are included in the systems through linear
patterning of the transparency substrate by laser printing lines on transparency. The adhesion energy of the PDMSheterogeneous substrate system is examined with respect to width and spacing of the linear patterns (i.e. 1mm,
2mm and 4mm), and orientation of the peel front with respect to the linear patterns (i.e., 0, 22.5, 67.5, and
90). During the peel test the peel front position, velocity and peel angle are monitored. It is observed that there
are oscillations in the peel force, and thus the adhesion energy, when the peel front travels from a region of higher
adhesion energy to a region of lower adhesion energy and vice versa.
Characterizing 7,795 Contact Binaries Using Survey Data
Franklin Marsh
Mentors: Thomas Prince and Ashish Mahabal
Contact Binaries are one of the most common types of eclipsing binary stars. In a contact configuration, two stars
are so close together, their photospheres touch. Until recently, only a small number of these systems have been
studied in detail. We aggregate data from the Catalina Real-Time Transient Survey, Palomar Transient Factory, and
Sloan Digital Sky Survey for the purpose of physically characterizing a large number of contact binary systems. Our
sample of nearly 8,000 systems allows the contact binaries to be studied as a population. We find that there is an
overabundance of binaries with photospheric temperatures of 6200 to 7000K, which have low mass ratios. We
show that starspots are the cause of the vast majority of contact binary light curve asymmetries (the OConnell
effect). Our sample is able to better define the period-temperature relationship for contact binaries, as well as
provide constraints on period change rates. Our project is an example of the science that can be performed with
publicly accessible survey data. In the future era of large, open access surveys, researchers will collaborate
globally, vastly expanding our knowledge of eclipsing binaries and other stellar systems.
An Information-Theoretic Perspective on Fluctuation Relations in Nonequilibrium Statistical Physics
Timothy Maxwell
Mentors: John Preskill, Ning Bao, and Nicole Yunger Halpern
Recent advances in nonequilibrium statistical physics have given physicists powerful tools for studying systems far
from equilibrium. We explored two recently discovered inequalities in nonequilibrium statistical physics: a
fluctuation relation by England from 2013, and a generalization called the "Bayesian Second Law of
Thermodynamics" by Bartolotta et al. in 2015. These inequalities put a lower bound on the entropy produced by a
physical process. We used analytic methods to show that England's result can be derived from the Bayesian Second
Law under certain assumptions. We used numeric methods to find that the Bayesian Second Law often provides a
tight lower bound on the entropy produced by the physical process in question; England's result provides a looser
bound than the Bayesian Second Law, but it is easier to apply. Our work will give the scientific community a better
understanding of these newly discovered fluctuation relations.
Pattern Order in Photoelectrochemically Grown Se-Te Films in Response to Light Intensity and
Deposition Time
Richard May
Mentors: Nate S. Lewis and Azhar Carim
Current photolithography faces difficulties in generating subwavelength features due to the diffraction limit and in
producing intricate three-dimensional structures due to the need for multiple masking and deposition cycles.
Recently, photoelectrodeposition of Se-Te semiconductor films has been shown to enable the growth of periodically
patterned structures as a function of illumination conditions. Photoelectrodeposited films were generated with a
series of illumination intensities (5 to 50 mW cm-2) as well as series of deposition times (30 to 300 s). Film
morphology was assessed via scanning electron microscopy. Fourier analysis was utilized to determine the average
lamellar periodicity in the resultant films as well as magnitude of the distribution of periodicities about this mean.

53

The lamellar periodicity was found to be independent of the illumination intensity but found to scale proportionally
with the deposition time. The distribution of lamellar periods was also found to be independent of the illumination
intensity but was found to tighten with increasing deposition time. Additionally, contrast thresholding of the SEMs
was used to determine the average fraction of film unpatterned due to dark growth processes during the
deposition. The unpatterned area was found to decrease with increasing light intensity and be mostly eliminated
after a threshold intensity.
Optimizing Self Organizing Map Templates for Determining Galaxy Color Distributions and Photometric
Redshifts
Daniel McAndrew
Mentors: Peter Capak and Daniel Masters
The elusive nature of dark matter, dark energy, and the large-scale 3-dimensional structure of the Universe remain
unknown, but they are prime targets of astrophysical investigation. Upcoming weak gravitational lensing
experiments from the \textit{Euclid}, \textit{WFIRST}, and other large surveys aim to map the geometry of the
Universe to demystify these monumental enigmas. To do this, the astrophysics community must calibrate
photometric redshifts for $\gtrsim 10^9$ galaxies. To deal with these seemingly daunting calibrations, we consider
a machine learning technique, the self-organizing map (SOM) algorithm. The SOM is a neural network that maps
the distribution of data in a high-dimensional space onto a lower-dimensional representation. We use the SOM to
characterize the empirical distribution of galaxies in color space, since a robust mapping of this distribution will help
develop a color-redshift relation for calibrating photometric redshifts. To improve the performance of the SOM, we
optimize a template set of galaxy spectral energy distributions (SEDs) on which to train the SOM. This is done by
iteratively optimizing ``eigenspectra" obtained by using the statical method of Karhunen-Loeve (K-L)
transformations on photometric data. We also introduce an error scaling factor to our implementation of the SOM
algorithm, to reduce the influence of faint objects with large photometric errors, mitigating the error and improving
the smoothness of the map.
Describing the Spatial Distribution of the Urban Built Environment With Networks
Sean McKenna
Mentors: Steven Koonin, Kaan Ozbay, and R. Michael Alvarez
Though the population and geographic extent of cities varies greatly, some urban experiences are ubiquitous, such
as crowded streets or closely-packed coffee shops. To a scientist, this suggests scaling laws inherent to urban
growth. Recent work has investigated the extent to which scaling is present in cities, investigating metrics of
economic growth, public safety, and infrastructure investment. This study investigates urban pedestrian networks;
it looks for scaling relationships in the spatial distribution of an urban areas built features across its network.
Specifically, distance relationships for stores, amenities, and emergency services are presented for several United
States metropolitan areas. Additionally, further characteristics of the distribution of real estate in New York City are
presented. Finally, an attempt will be made to show that a marriage of the monocentric city model with standard
land-developer assumptions of building characteristics can predict the geographic extent of metropolitan areas.
The Effect of Sleep on the Avoidance Response of C. elegans
Katelyn McKown
Mentors: Paul Sternberg and Ravi Nath
Sleep is important for survival and health, yet little is known about human sleep. Because many genes are
homologous between Caenorhabditis elegans and humans, characterizing C. elegans sleep can lead to a better
understanding of the genetic basis of sleep and perhaps lead to better treatment for sleeping disorders. Stress by
heat shock can induce a quiescent, sleep-like state in C. elegans. ALA, a neurosecretory cell, is necessary for this
sleep-like state. Of the known characteristics of sleep, I have focused on studying the response latency of
C. elegans to 1-octanol, an aversive stimulus. To accomplish this, I have conducted avoidance assays on awake
and sleeping C. elegans. These series of experiments have allowed me to determine the dose-response curve of
C. elegans to 1-octanol. Additionally, I performed in vivo calcium imaging experiments on sensory neurons to
further study the activity of C. elegans neurons in awake and sleeping animals.
Effects of Menthol and Nicotine on the Upregulation and Stoichiometry of 6* and 4* nAChRs
Emily Meany
Mentors: Henry Lester and Brandon Henderson
Neuronal nicotinic acetylcholine receptors (nAChRs) are ion channels formed by pentamers of (2-10) and (2-4)
subunits which regulate the release of acetylcholine, dopamine, and other neurotransmitters. At the level present
in tobacco products, the agonist nicotine upregulates high sensitivity 4(223 and 623 nAChRs. In addition to
nicotine, about 90% of cigarettes contain menthol flavorant, which has two primary stereoisomers (denoted plus
and minus). As both forms are found in menthol cigarettes depending on manufacturing process, we are interested
in any differences in effects between them. Neuro-2a cells and mouse brain slices were used to investigate changes
in stoichiometry and upregulation due to each stereoisomer alone and in combination with nicotine. Stoichiometry

54

was investigated using Normalized Frster Resonance Energy Transfer (NFRET) while nAChR export to the plasma
membrane, a step in upregulation, was studied with quantification of ER exit sites (ERES). Neuro-2a cells were
transiently transfected with fluorescently tagged 4 subunits for NFRET and ERES experiments and mice containing
fluorescent 4 (mCherry) and 6 (GFP) nAChR subunits were used for in vivo NFRET assays. We expect that the
plus form of menthol will be less potent than the minus form and that menthol will upregulate the less sensitive
432(2 and 62(non-3) forms, which would support previous data and could contribute to habituation and
tolerance in combination with nicotine.
Examining the Transition Into Many-Body Localization in One Dimensional Quantum Spin Chains
Alexander Meiburg
Mentor: John Preskill and Jenia Mozgunov
We consider the Hamiltonian given by H(s) = si xi + (1-s) (i Ji zi zi+1 + i hi zi zi+1) on a 1D quantum spin
chain. The Ji's and hi's are drawn randomly and independently from the uniform distribution on [-1,1]. At s=0, this
is a set of non-interacting spins with a x-oriented external field; at s=1, it consists entirely of z-interaction terms at
random on-site potentials. It was previously known that for intermediate s (~ 0.5) this has spin waves that
propagate and non-zero heat conductivity, whereas at s=1 it enters a many-body localized phase and there are no
travelling waves. We try to characterize the transition from this extended phase to localized phase through a
variety of parameters about the system, including ground state energy gap, magnetic susceptibility, von Neumann
entanglement, a level-clustering parameter r, and behavior under adiabatic evolution.
Characterization of Water-Jet Induced Gas Ionization by Optical Emission Spectroscopy
Sean A. Mendoza
Mentors: Mory Gharib and Francisco Pereira
A High-speed distilled water-jet, shown to generate regions of air excitation, is characterized with optical emission
spectroscopy techniques. It is determined that the excitation energy of the external charge on a neat water jet
produces an ionized region of air around the area of impingement on dielectric crystal materials. The ionized region
of air emits light spectra identical to that of an electric air corona. When air is replaced with other gases, an
equivalent ionization spectra is observed. When the gas environment is changed to a gas with a higher ionization
energy however, gas ionization is instead replaced by non-ionizing gas excitation. Mechanisms of action for waterinduced air ionization are discussed.
Finding Stationary Solutions to the Chemical Master Equation by Gluing Graphs at One or Two Vertices
Xianglin (Flora) Meng
Mentors: Richard M. Murray, Ania A. Baetica, and Vipul Singhal
Biological systems display significant stochastic behavior at low molecular numbers, which requires stochastic
modeling. The chemical master equation (CME) is a stochastic model that depicts a continuous-time Markov
process whose state space comprises possible combinations of the numbers of molecules of each species. However,
stationary solutions to the CME are generally unknown. Mlykti et al. recently proposed a probability technique
that derives the stationary distribution of a continuous-time Markov process whose state space can be constructed
by gluing two finite, irreducible state spaces at one or two vertices. We study the applicability of the technique to
finding stationary solutions to the CME. We find the set of reactions whose stationary distributions are obtainable
using the technique, which includes reactions whose state spaces are trees of trees and cycles. In addition, we
propose a recursive algorithm that minimizes the computation cost when the technique is applied to closed reaction
systems with three species. In particular, closed systems with three species and two reversible monomolecular
reactions have state spaces that are triangular grids, and we derive explicit expressions for their stationary
distributions. Knowing the stationary solution to the CME will help inform experimental design as well as the design
of distributions.
Binding Interactions of the ATP-Independent Disaggregase cpSRP43 With Its Substrate
Emily Miaou
Mentor: Shu-ou Shan
Membrane proteins contain hydrophobic regions that cause them to be prone to aggregation in aqueous
environments, resulting in difficulties maintaining a fully-functional post-translational state as they move through
aqueous cellular compartments. Novel chaperone activity has been identified in the chloroplast signal recognition
particle (cpSRP), which contains a subunit cpSRP43 that has been found to be capable of preventing the substrate,
light-harvesting chlorophyll-binding proteins (LHCPs), from aggregating in solution as well as resolubilizing
aggregated proteins in vitro using extensive binding interactions rather than ATP. This project sought to better
understand the interaction between cpSRP43 and LHCP by mapping the binding sites of LHCP on cpSRP3. Potential
binding residues on cpSRP3 were identified, site-directed mutagenesis was performed to introduce single-cysteine
mutations, and protein expression and purification were done to obtain the mutant chaperones. Light-scattering
assays and anisotropy showed that many of the mutants had defective binding with LHCP, but a few neutral
mutants were identified that will be used for future alkylation-protection experiments. Additionally, the activity of

55

some soft mutations may be restored by inserting the mutations into a superactive mutant, thus allowing
alkylation-protection experiments to be performed on those mutations as well. Mapping the cpSRP43-LCHP
interaction helps create an important model for understanding how cells process aggregation-prone membrane
proteins, such as those involved in neurodegenerative diseases including Alzheimers, Parkinsons, and
Huntingtons.
Mechanical Behavior of Three-Dimensional Tensegrity Nanolattices
Talia Minear
Mentors: Julia Greer and Arturo Mateos
Tensegrity systems are lightweight structures that exhibit a high weight-to-strength ratio due to an efficient
distribution of mass and force among its structural elements. Literature exists on macroscopic tensegrity systems,
but very little exists on the mechanical behavior of tensegrity structures at the nanoscale. We investigate a new
class of structural metamaterials composed of tensegrity architecture, which have the potential to enhance
mechanical properties and simultaneously take advantage of size effects at the nanoscale. Minimal, regular,
tensegrity lattices made out of polymer were fabricated using two-photon lithography and various arrangements of
the lattice unit cell were fabricated. Compression experiments were performed and the mechanical properties and
deformation of these nano-tensegrities were studied.
Probing Quasar Physics With Variability and Machine Learning
Catalina-Ana Miritescu
Mentors: Matthew Graham and George Djorgovski
Quasars are cosmic objects which have long been known to be variable sources. Their variability is being monitored
by digital sky surveys, the largest open time domain survey currently operating being the Catalina Real-time
Transient Survey (CRTS). A new variability-based method has been developed to analyze variable objects: the
Slepian Wavelet Variance (SWV). The SWV is a statistical characterization of a large set of time series (light
curves). In this project, we used SWV on a subset of CRTS data to determine the general trend of each known
subtype of quasars (classes determined spectroscopically) and to identify possible outliers. From the outliers
identified, we selected as objects warranting further study the ones whose light curves had unusual appearance.
Using the specific behaviors for each class determined from the subset, we can extend the search for objects of
interest to the whole CRTS database.
Development of Biligand Capture Agents to Target Plasmodium falciparum Histidine-Rich Protein II
(PfHRP2) for Rapid Malarial Diagnosis
Anvita Mishra
Mentors: James Heath and JingXin Liang
Currently, malarial rapid diagnostic tests use antibodies to detect Plasmodium falciparum histidine rich protein II
(PfHRP2), a biomarker for Plasmodium falciparum malarial infection. Malaria caused by this species is especially
deadly and has the highest rate of fatality. However, these antibodies are too costly and too thermally and
chemically unstable to be used viably in third world countries. Thus, the Heath lab is developing an antibody-free
rapid diagnostic test for the diagnosing malaria by developing protein capture agents which can target the entire
PfHRP2 protein.
Macrocylic peptides against distinct epitopes of PfHRP2 have been developed. Nevertheless, we are aiming to
develop capture agents with higher binding affinity than the monoligands in order to develop accurate tests which
have a low limit of detection. My work involves increasing the binding affinity of the capture agents by linking two
PfHRP2 monoligands to create biligands. We want to test a cooperativity hypothesis to see if binding multiple sites
simultaneously with an optimized linker will yield better affinity. The PfHRP2 binding affinity of the biligands was
evaluated and compared to the monoligands through colorimetric enzyme-linked Immunosorbent assays (ELISAs).
This report describes in detail the procedure to synthesize these biligands, along with the process to find the best
biligand binders for PfHRP2.
Investigation of Neuronal Populations Controlling Thirst
Jisoo Mok
Mentors: Yuki Oka, Vineet Augustine, and Nikki Cruz
The fluid homeostasis is the set of processes that regulate the salt and fluid balance of the body. When this balance
shifts, several regions in the circumventricular organs (CVO) of the hypothalamus are activated, and the
stimulation of specific brain regions can lead to changes in drinking behavior. The Oka Lab is currently investigating
the subfornical organ (SFO) and the organum vasculosum of lamina terminalis (OVLT) to discover a molecular and
genetic mechanism behind this brain-body interaction that controls the fluid homeostasis and hence influences that
drinking behavior. Although the Oka lab has recently demonstrated that the two regions mentioned above govern
the balancing of the fluid homeostasis, it is still unclear which neuronal population is involved. The lab has been
utilizing the optogenetics and immunohistochemistry to find the neurons that trigger or sense the changes in the

56

fluid homeostasis and to understand the genetic background. One of the neuronal markers, ER81, which could be
tied with the fluid homeostasis, has been identified with the immunohistochemistry technique. Further investigation
of the SFO and the OVLT and search for new neuronal markers will offer better understanding of how the mouse
brain governs the drinking behavior and the fluid homeostasis.
Development of Tunable Mammalian Synthetic Transcriptional Regulators for Biomedical Applications
Andrew Montequin
Mentors: Ahmad Khalil and Michael Elowitz
Previous research has led to the development of a new class of synthetic transcription factors (sTFs) based on
programmable zinc finger domains to precisely control gene expression and enable orthogonal recruitment of
proteins to site-specific locations in the yeast genome. The goal of this project is to advance these tools and
frameworks for logically controlling genes in mammalian systems for next-generation biomedical applications. To
this end, we developed a catalog of synthetic mammalian regulatory elements, which we used to create novel zinc
finger-based sTFs. We characterized the ability of these sTFs to regulate reporter genes driven by engineered
promoters in HEK293 cells and explored transcriptional outputs for a variety of different transcriptional activation
domains and DNA binding affinities, which we modulated via mutagenesis of conserved residues in the zinc finger
backbone. Ongoing work is being done to engineer the conditional expression of these novel sTFs based on
endogenous signals, such as hypoxia or [T-cell activation/signaling], which could hold promise for genetic tools for
cell-based therapy.
Vacuum Technology for Applications in Optoelectronics and Novel Environments
Aadith Moorthy
Mentors: Axel Scherer and William M. Jones
Novel miniaturized vacuum triodes and vacuum-based photodetectors, both of which utilize field emission of
electrons, were studied in this research work. The nano-scale vacuum triodes were fabricated using advanced
lithography and their current-voltage characteristics were studied in detail, with emphasis given to their
temperature variation and switching speed. The novel photodetector was proposed and analyzed computationally
using the finite element method to discern suitable geometries for its most efficient function. The vacuum triodes
fabricated and tested by these methods showed exponential current-voltage curves and when subject to
temperature differences, showed an increase in emission with increasing temperature, as predicted by theory. They
could also be switched efficiently using an oscillator circuit, showing their promise in future electronics. The
photodetector computation model showed that an optimum field emission-based photodetector could be built using
a slotted waveguide that had small teeth on the sides of width 60nm, height 40nm and separation 70nm. In
addition, taking advantage of Surface Plasmon Polaritons (SPP) by placing small metal beads at the tips will enable
efficient electron emission and subsequent photodetection. Overall, the results substantiate the promise of field
emission in vacuum technologies for novel applications and environments.
Stability of the Immersed Boundary Lattice Green's Function Algorithm
Ben Morley
Mentors: Tim Colonius and Sebastian Liska
The Immersed Boundary Lattice Green's Function (IBLGF) algorithm is a new algorithm being developed at Caltech
for simulating incompressible flow about an arbitrarily-shaped obstacle. The partial differential equations governing
the flow are discretized in space to form a set differential algebraic equations (DAEs) that enforce the NavierStokes equations under the constraints of incompressibility and the no-slip boundary condition along the immersed
surface. We show that the numerical solution of the linearization of these DAEs does not require significant
additional stability conditions beyond those arising from the unconstrained equations. Using this and an analysis of
the semi-discrete equations we argue that, in most situations, similar stability conditions are likely valid for the
non-linear equations. The analysis presented is not specific to the IBLGF algorithm; it shows the stability of a
commonly encountered class of linear DAEs in a manner that does not depend on any one choice of integration
scheme, but instead applies to a range of projection based schemes.
Efficient Analysis of Sensitivity in Nucleic Acid Systems to Model Imperfections
Rachael Morton
Mentors: Niles Pierce and Nicholas Porubsky
The ability to program the pairing of nucleic acid bases is useful for many different applications, particularly for
engineering self-assembling molecular structures and systems. Much preliminary work has been done in analyzing
how nucleic acid strands interact, and in designing equilibrium base-pairing properties of an interacting nucleic acid
strand complex. An algorithm for designing these latter properties has been developed. A key component of this
algorithm is hierarchical ensemble decomposition: an ensemble is decomposed into leaves, where calculations
are made, and merged upward until the original root level is reached. This study first involves testing the
algorithm to examine how predictions alter when parameter values are perturbed. This was accomplished by
perturbing random and engineered nucleotide sequences, and analyzing outlier sequences that produced results

57

deviating from what was expected. Next, this study investigates the effectiveness of using a less computationally
intensive technique of estimating root level properties from leaf level calculations, and analyzes why certain
nucleotide sequences yield incorrect predictions, and what components in the design algorithm cause this.
Consistent error trends suggest a flaw in the algorithm, but show that utilizing leaf level calculations produces
reliable results. Further investigations will examine possible indicators for sequences with poor leaf level
calculations.
Locating Primordial Stars in the Outskirts of Nearby Dwarf Galaxies
Eric Roger Moseley
Mentor: Evan Kirby
The dwarf spheroidal galaxies Sextans and Ursa Minor orbit the Milky Way. Many dwarf galaxies exhibit radially
decreasing gradients in their stellar metallicities. As a result, it makes sense to quantify the metallicity gradients in
these galaxies and look in their outskirts for exceptionally metal-poor, primordial stars. The spectra of about 1000
stars total in the line of sight of these dwarf galaxies were analyzed for the quality of their spectra. Provided the
spectra were good, we found the metallicities and heliocentric radial velocities of these stars and attempted to use
these to determine their probabilities of galactic membership. In addition, if the stars lay too far from a best-fit
color-magnitude isochrone, their membership probability was set to zero. For all but about a dozen stars, we have
so far determined at the 99.99% significance level their galactic memberships and have found within error their
metallicities. A few relatively distant and metal-poor stars had their spectra taken with the DBSP spectrograph on
the Hale telescope, but this data has not been analyzed yet. More stars from the members may still be identified as
good candidates for research into first generation, metal-poor stars.
Examining the Functional Differences of Drosophila FGFs Pyr and Ths by Examining in vivo Localization
of Ths and Testing a TMD in Pyr
Zane Murphy
Mentors: Angelike Stathopoulos and Vincent Stepanik
In Drosophila, two Fibroblast Growth Factors (FGFs), Pyramus (Pyr) and Thisbe (Ths) help guide several coordinate
cell migrations during embryogenesis. This includes formation and migration of the Caudal Visceral Mesoderm
(CVM), a group of cells that establish the longitudinal visceral muscles. These two FGFs are closely related and
overlap functionally, but are both required for efficient CVM migration. One structural difference that may
contribute to this functional difference is a putative Transmembrane Domain (TMD) in amino acids 400-425 of Pyr
which is not present in Ths. Using Splicing by Overlap Extension (SOE) PCR, an expression construct for a fusion
protein was created from the native signal peptide of Pyr and its putative TMD ligated into fluorescent markers to
test if the TMD is sufficient to localize proteins to the cell membrane in cell culture through transient transfection.
To further explore the properties of Drosophila FGFs in vivo, antibody staining was performed on Drosophila
embryos to visualize the localization of a mCherry-tagged Ths fusion protein under control of its endogenous
regulatory sequences, as the distribution of Pyr and Ths have not yet been described. Using these data in
correlation with the known localization of CVM cells could provide new insight into the role of each FGF in this
process. This study also has general implications, showing that differences in evolutionary development of signaling
molecules can lead to more severe functional and structural disparities.
Raman Spectroscopy of OH Vibrational Modes in Tourmaline Crystals
Xinyi Nan
Mentor: George R. Rossman
The OH vibrations of tourmaline were studied using polarized and unpolarized Raman spectroscopy as a potential
proxy for investigation of the metal cation sites. The sample set consisted of 25 elbaites from 12 localities, 15
dravites from 11 localities, 8 liddicoatites from 3 localities, and 4 uvites, 3 schorls, 2 buergerites, 2 rossmanites, 2
foitites, 1 darrellhenryite and 1 olenite, all from different localities. Curve fitting of each polarized Raman spectrum
revealed between 5 and 10 distinct bands in the region 3300 cm to 3700 cm
. T h o u g h b an d s varied
number, energy, intensity, and shape depending on the sample, consistent features were found within species for
elbaite, dravite, uvite, schorl, buergerite, and foitite. Similarities were also found between elbaite and rossmanite,
as well as between dravite, uvite, and schorl. The OH spectral features of buergerite and olenite were significantly
less intense than those of the other species studied, indicating a rough quantitative correlation with hydrogen
content. These results coupled with FE SEM analyses suggest that Raman spectroscopy of the OH region alone
could be used to distinguish between some tourmaline species and give valuable insight into chemical composition.

58

Electrical Impedance Tomography as a Novel Method of Assessing Vulnerable Plaques

Anusha Nathan
Mentors: Yu-Chong Tai, Shell Zhang, and Yuan Luo
Rupture of artherosclerotic plaques is the most common mechanism through which unstable angina, acute
myocardial infarction, and sudden cardiac death occur. Current methods of locating these plaques include X-ray
angiography and intravascular ultrasound, both of which are expensive and lack resolution and penetration depth.
Electrical impedance tomography exploits the dielectric properties of biological tissues to distinguish lipids, i.e.
plaques, from other tissues in the blood vessels. Using Flexible Flat Cable, electrode rings with 32 channels were
fabricated to conduct inward and outward-facing EIT in a phosphate-buffer solution and in healthy and plaqueinfested pig aortas. Swisstom software was used in conjunction with MATLAB EIDORS EIT algorithms to
successfully image plaque-infested, pig aortal walls. The signal-to-noise ratio of the devices was measured to
determine the efficiency of this specific implementation of EIT. These ex-vivo experiments serve as prototypes for
in-vivo testing using a delivery catheter to insert a similar microelectrode device into rabbits to detect plaques on
their aortal walls.
Analysis of Neocortex Tissue Derived from Induced Pluripotent Stem Cells to Study Pathogenesis of
Human Lissencephaly
Aishwarya Nene
Mentors: Arnold Kriegstein, Marina Bershteyn, and Marianne Bronner
The human cerebral cortex is an immensely complex structure whose critical functions are disrupted in
developmental disorders, such as human lissencephaly. Classical lissencephaly is a rare genetic disorder
characterized by a smooth cerebral surface, mental retardation, and seizures. Current understanding suggests that
lissencephaly is caused by defective neuronal migration, but there has been no human model to study
lissencephaly. This project uses reprogrammed induced pluripotent stem cells (iPSCs) from healthy and
lissencephaly patient fibroblasts to study early cortical development using three dimensional organoid cultures. The
differentiated tissue was analyzed by time-lapse imaging and immunohistochemistry (IHC) to study neuronal
migration. Consistent with previous data from mouse models, lissencephaly neurons were significantly deficient in
migration, validating this novel model system. Furthermore, IHC analysis was used to characterize the cleavage
angles of lissencephaly ventricular neural progenitors, namely radial glia cells. These studies revealed an increase
in oblique and horizontal divisions, suggesting premature neurogenesis. Results from studies on human 3-D models
of lissencephaly will help elucidate disease mechanisms, which may lead to the development of new therapeutic
strategies.
Gas Vesicles as Functional Ultrasonic Reporters
Suchita Nety
Mentors: Mikhail Shapiro and Anupama Lakshmanan
Ultrasound is a widely used non-invasive imaging modality in biomedicine, yet its high spatiotemporal resolution
has not yet been fully exploited for molecular imaging due to the lack of suitable contrast agents. Gas vesicles
(GVs), hollow protein-shelled nanostructures from buoyant microorganisms, have recently been identified as a new
class of ultrasonic reporters, and our present goal is to engineer GVs as functional imaging agents. We aimed to
enable covalent attachment of any protein of interest to the GV surface by employing the genetically-encoded
SpyTag-SpyCatcher protein-tagging system. SpyTag-coated GVs were prepared and reacted with fluorescent
SpyCatcher proteins in vitro. Collapse pressure assays and ultrasound imaging indicated that covalent attachment
of fluorescent proteins to the GV surface did not perturb GV mechanical and acoustic properties. GV fluorescence
was confirmed by fluorescence microscopy, and the degree of labeling was quantified by SDS-PAGE. Ongoing
efforts are focused on co-labeling GVs with multiple proteins (i.e. targeting peptide and fluorescent protein), thus
establishing GVs as functional multimodal contrast agents.
Optimizing Synthetic Capture Agent Targeting G12D Epitope of Oncoprotein KRAS
Rachel Ng
Mentors: James R. Heath and Ryan K. Henning
Mutations in GTPase KRAS are frequently found in human cancers. In particular, G12D is the most frequent
substitution mutation that disables GTP hydrolysis in oncoprotein KRAS. The GTP-bound KRAS remains functionally
active and activates cascades of intracellular pathways that propagate cancer. Here, we employed in situ click
chemistry to develop cyclic protein capture agents that not only have high binding affinity and selectivity for
mutant KRAS(G12D), but also are stable chemically, biochemically, and thermally. Enzyme-linked immunosorbent
assays at different KRAS concentrations produced saturation curves that provide EC50, a quantification of the
capture agents binding affinity for wild type and mutant KRAS. Further medicinal chemistry is done to optimize the
capture agents binding. We also conducted alanine scans, which determine which amino acids are necessary for
binding or changeable. Additionally, we tested fluorinations at different carbon positions in the capture agent to
increase the capture agent binding and lipophilicity.

59

Design Space Exploration of the Violacein Pathway in Escherichia coli Based Transcription Translation
Cell-Free System (TX-TL)
Phuc H.B. Nguyen
Mentors: Richard Murray and Yong Y. Wu
Violacein is a violet, pyrrolidone containing compound with anti-cancer, anti-fungal, anti-viral, and anti-oxidation
properties. The pigment is natively synthesized by Chromobacterium violaceum. Nevertheless, its production is too
expensive due to the strains low productivity. Reconstructions of this pathway in heterologous hosts have been
done to improve the yield. The approach, however, suffered from buildup of intermediates and byproducts due to
flux imbalances. In this study, Escherichia coli Based Transcription Translation Cell-Free System (TX-TL) was
employed to sample various expression levels of the violacein pathway. TX-TL enables rapid modifications and
prototyping of the pathway without complicated cloning cycles. Linear DNAs of the violacein pathways enzymes
with 100 bps protection flanking sequences were expressed in TX-TL. The products were collected in different
solvents and analyzed via UV-Vis absorption and LC-MS. Violacein production at different linear DNA concentrations
of each enzyme were explored. We demonstrated that the violacein metabolic pathway has been successfully
reconstructed in TX-TL. Different variants of the pathway were shown to produce different colored products.
Violacein synthesis was confirmed using LC-MS. Design space exploration suggested a high expression of VioC and
VioD would lead to an improved violacein production level. This result can be used for the future engineering of
more efficient violacein producing E.coli strain.
Ramified Coverings of the Complex Projective Line With Semisimple Galois Groups
Tian Nie
Mentor: Dinakar Ramakrishnan
Let 1 be the complex projective line and C a smooth projective algebraic curve. Suppose that f: C 1 is a ramified
covering of 1 with ramification locus in {0,1, }. Note that the fundamental group of 1 {0,1, } is , the free
group of two generators. Hence any quotient of this free group can be realized as the Galois group of some curve
C. Consider the Heisenberg group which is isomorphic to the matrix subgroup of 3 (/) generated by
1 1 0
1 0 0
0 1 0 and 0 1 1. We studied the defining equation of a curve which has as its Galois group. We are
0 0 1
0 0 1
also interested in the cusps of .
Thermoelectric Power Generation From Waste Heat for Wireless Aircraft Sensors
Nirmal Jayaprasad
Mentors: Austin Minnich and Navaneetha Ravichandran

Thermoelectric devices can recover waste heat generated in jet engines and convert them into electricity. In this
project, we design and fabricate a wireless sensor prototype powered by a thermoelectric generator (TEG), which
can measure temperature, pressure and air quality in aircrafts. When a temperature gradient is applied across the
TEG, it generates power to charge a rechargeable Lithium-ion battery. An LTC3108 power management circuit
boosts the output voltage obtained from the TEG to meet the sensor power requirements and an over-charge
protection circuit protects the battery and prevents unnecessary discharge in the absence of thermal gradients. The
weight and size of the prototype was minimized since they pose critical issues in aerospace applications. The final
prototype works for various temperature gradients as low as 400 C and the TEG successfully powers the wireless
sensor which transmits data wirelessly over a network connection. Hence, this TEG-powered sensor is capable of
being used in commercial aircrafts at remote locations where electric power is not readily accessible.
Identifying and Characterizing TCRs From Elite Controller T Cells
Won Jun Noh
Mentors: David Baltimore and Alok Joglekar
Current methods of HIV therapy, such as the use of antiretroviral drugs, are insufficient in fully suppressing HIV, as
improvements must be made to effectively control latently infected cells and newly created drug-resistant species.
Further investigation of HIV elite controllers, a rare subpopulation of HIV-positive patients known to be able to
achieve broad control over HIV without any separate treatment, may contain the key to finding a potent, longlasting treatment for HIV infections. Recent studies on elite controllers have shown that the controller status is
related to CD8+ Cytotoxic T Lymphocytes (CTLs) that are restricted by certain HLA-B alleles. Our collaborators at
Bruce Walkers laboratory have shown that certain B27-KK10-specific CTL clonotypes from elite controllers exhibit
higher potencies in eliminating HIV-infected cells. Considering that the only genetic differences between these
clonotypes lie in T cell receptors (TCRs), we expect that controller status is related to the TCRs. In this study, we
isolate and characterize elite controller TCRs by amplifying TCR/TCR transcripts from the RNA extracted from
B27-KK10-specific elite controller CTLs. We tested TCR surface expression levels through flow cytometry that
utilizes B27-KK10-specific tetramers. TCRs that are able to recognize B27-KK10 will be studied further to test their
abilities to control HIV.

60

The HED Meteorite Clan: Visible-to-Near-Infrared Spectral Diversity at 81m Spatial Resolution
Geraint Northwood-Smith
Mentors: Bethany Ehlmann and Abigail Fraeman
The HED meteorite clan are mafic and ultramafic igneous rocks and impact breccias, widely thought to originate
from the asteroid 4 Vesta. Eucrites are surface basalts or shallow cumulate gabbros, while diogenites formed in
plutons at greater depth. Howardites are a regolith breccia of eucrite, diogenite and exogenous material, possibly
including olivine. In 2011 the Dawn spacecraft provided visible-to-near-infrared spectra of the Vestan surface with
70m per pixel resolution, however this resolution is insufficient to reliably extract all mineral components. In this
study, a collection of HED meteorites are analysed using the Ultra-Compact Imaging Spectrometer (UCIS) at
approximately 81mm per pixel resolution. Methods of hyperspectral characterisation are developed to reveal their
spectral diversity at this small spatial scale and their mineralogy is interpreted. Using these methods, a variety of
mineral classes has been distinguished in the samples including olivine, feldspar, multiple classes of pyroxene, and
glass/melt material. Spectral variation within these classes and between lithologies has also been characterised.
Results from this study will inform interpretations of Vestan surface spectra, and provide evidence regarding the
cooling history of Vesta, the identification of exogenous material and its interaction with native lithologies, and
specifically the proportion of olivine that is exogenous.
Genome Editing in Sea Urchins: Developing CRISPR/Cas Systems for Strongylocentrotus purpuratus
Ariel Margaret ONeill
Mentors: Eric H. Davidson and Miao Cui
Clustered Regularly Interspaced Short Palindromic Repeats (CRISPR) and CRISPR-associated (Cas) proteins
comprise an adaptive immune system first discovered in Streptococcus thermophilus. The CRISPR/Cas system
allows the bacteria to edit its own genome in order to develop resistance to a virus. Researchers have harnessed
the genome editing potential of CRISPR/Cas to create transgenic lines of bacteria, yeast, mice, worms, zebrafish,
and human tissue. However, the technique has not yet been adapted for use in sea urchins, a key model organism
for the study of gene regulatory networks. This project is part of a continuing effort to develop a reliable and
efficient method of generating transgenic lines of sea urchins.
A Determination of an Explicit Set of Coset Representatives for Double Coset Spaces Involving
Open-Compact Subgroups of SO5
Tynan Ochse
Mentor: Pei-Yu Tsai
The Theory of Automorphic Forms is nearly ubiquitous in modern Number Theory. Studied originally for the
connection with elliptic curves, the rich interplay of mathematics that takes place did not reveal itself until the
1960s, and in particular, with the formulation of the Langlands Program. Recently, a local newform theory has been
developed by Pei-Yu Tsai for Split Special Orthogonal Groups of Odd degree. In order to understand the higher
order cases, one first needs to understand the rank 3 case. The purpose of this paper is to determine a set of coset
representatives for SO5 and SO7 by first computing the corresponding root system, which then can be used to
deduce the structure of the Weyl Group, W. This information is key in determining the Bruhat global decomposition
for these two algebraic groups. From there, the local newform theory for SO2n+1 is used as a guide to work out a
suitable set of coset representatives for particular double coset spaces. This can then be used to study the
generalized Hecke Algebra for open-compact subgroups of SO5 and SO7, which in turn provides arithmetic
information on the coefficients of the corresponding automorphic forms.
Immersive and Collaborative 3D Data Visualization
Sunwoo Oh
Mentors: George Djorgovski, Ciro Donalek, and Scott Davidoff
With massive amount of data, it is becoming more and more important to find patterns hidden in it. Data
visualization is a effective way to find that pattern, which is also coherent with human intuition. However, in
visualizing the data in immersive 3D environment, we use data points in different shape classes and sizes. There
are many factors to consider, such as lighting, perspective, color, shape, displacement that affect the perceived
size of the objects. Therefore, I tried to make a standard for objects, or functions among the shapes that indicate
the equivalent size of the objects. My program is made using Unity 3D based on C#, for users to scale the size of
one of given objects compared to another fixed object, both objects given randomly in color, shape, displacement.
The scaling should be done using Leap Motion and Oculus VR in immersive environment, to examine the effect of
being in immersive mode correctly. I also studied further about general big data analysis and data mining, since
the ultimate purpose of the project is in that context.

61

Synthetic Studies Towards the AEF Ring System of Talatisamine

Ciara Ordner
Mentors: Sarah Reisman and Victor Mak
Talatisamine is a complex, hexacyclic, C19-diterpenoid alkaloid that selectively blocks potassium ion channels.
Thus, it could be used as a potential treatment for Alzheimers disease. In addition, talatisamine, as well as other
diterpenoid alkaloids, have a variety of significant biological activities. The ultimate goal of this project is to
develop an efficient synthetic route to talatisamine. The first step toward this goal is to access the AEF tricycle.
Studies were conducted up to an epoxyketone intermediate of the proposed AEF tricycle synthesis. The five step
synthesis of an epoxyketone intermediate was carried out on both small and large scales, providing an ample
amount of material to explore subsequent chemistry. This project will continue in order to construct the AEF
tricycle.
Statistical Analysis of Experimental Data on Laminar Flame Speeds
Jeffrey Orenstein
Mentor: Guillaume Blanquart
This research seeks to perfect our pre-existingmodel for SL by re- analyzing pre-existing data with a more
statisticalapproach. We seek to characterize existing experimental measurements by their statistical and epistemic
uncertainties. We intend to use these uncertainties to show that the flame speed of all normal alkanes with at least
five carbon atoms can be calculated using one model. We will accomplish this by attempting to show that for a
given , T, and P, the uncertainties for all data points overlap. In addition, we will use the uncertainties we
determine to identify the most valuable future experimental measurements to perform, as well as to extract a new
function for SL (, T, P).
Randomized Methods for the Clustering Problem
Berk zdalyan
Mentor: Babak Hassibi
Graph clustering algorithms attempt to identify clusters from a noisy data source. We consider the Augmented
Lagrange Multiplier (ALM) method for the RPCA problem and use the Stochastic Block Model to generate pseudodata. In order to improve computation time without significant loss in accuracy, we apply randomized methods
which approximate the basis of the input data and compute singular value and QR decompositions. Our simulations
indicate that for sufficiently dense and large clusters, randomized methods can be successfully employed to recover
the underlying denoised structure. We present extensive simulations and suggestions for implementation. In
particular, for large size clusters with edge probability 0.7 or greater and a good regularization parameter, it can be
shown that randomized methods outweigh classical algorithms in performance without any loss in accuracy. These
results can be extended to the case of mid-size clusters for edge probability 0.85 or greater. In the future, we hope
to present our readers with rigorous theoretical bounds which justify the implementation heuristics.
Measurement of T Cell Receptor Mispairing for Selection of Effective Gene Therapy TCRs
Meghana Pagadala
Mentor: David Baltimore
The heterodimeric / T cell receptor (TCR) is the sole determinant of T cell specificity. In TCR gene therapy,
patient T cells are transduced with tumor-specific and TCR genes to impart anti-tumor immunity. However,
introduced TCR and chains can mispair with the transduced T cells endogenous TCR and chains,
respectively, reducing the number of tumor-specific TCRs on the surface and potentially generating autoreactive
TCRs. So far, a TCRs propensity to mispair has been an undefined property. We are developing a quantitative
assay to measure the extent to which mispairing occurs for TCRs of clinical interest. We identified two antibodies
that bind the constant portions of the endogenous TCR and chains. These epitope sites were mutated in the
transduced TCR chains and synthetic epitope tags were added to the N-terminus of these chains to enable their
orthogonal recognition by a second set of two antibodies. Using a sandwich ELISA, we capture the transduced or
TCR chain with a tag-specific antibody and then detect the or chain to which each is paired using a second
antibody. Signal is observed only when both capture and detection antibody targets are present. Therefore, all
possible / heterodimers can be distinguished by using different antibody pairs and each heterodimer can be
quantified by comparison to a standard curve. Although there were no clear differences in the mispairing ratios of
TCRs within a cell, there seems to be a stratification of mispairing ratios for TCRs only expressed on the surface.
Since the assay focuses on ratios, the variability is not due to TCR strength (how well a TCR is expressed on the
surface), but rather an inherent property of TCRs that determines whether a mispaired heterodimer will be
expressed. Future experiments will hopefully validate our results.

62

Psychological Phenomena in Financial Decision Making

Jay Palekar
Mentors: Andrew Lo and Antonio Rangel
Developments in behavioral economics have called into question theories that assume rational behaviors in
humans. We sought to test the validity of the Binary Choice Model, an evolutionary explanation of behavior, as an
alternative theory of decision-making. In order to do so, we carried out several types of behavioral studies on
individuals to understand their decision making process. In the first study, individuals were asked to make
decisions in both financial and medical contexts to measure the extent of risk aversion in their behavior, while in
the second study we asked individuals to guess two hundred times which of two images would appear on a screen
and studied how different rewards for guessing incorrectly motivated them to probability match. In particular, we
were interested in identifying demographic factors that contributed to individually irrational behaviors.
Mechanism and Kinetics of Substrate Binding to DNA-Labeled Surfaces
Lisa Pangilinan
Mentors: Jacqueline K. Barton, Ariel L. Furst, and Michael G. Hill
Cyclic voltammetry with a rotating gold-disk electrode (RDE) was used to examine the kinetics of methylene blue
(MB) binding and surface concentration of MB intercalated on DNA-modified electrodes. Data were fit to a Langmuir
model to determine k1 (adsorption) and k-1 (desorption) values. Small MB adsorption charge values indicate poor
DNA monolayer formation on the gold electrode surface. Further trials will be performed to determine and confirm
the kinetics of MB in various single base-pair mismatched DNA-modified electrodes. Redox reactions of intercalated
molecules bound to DNA-modified films are an important part in understanding charge delocalization and base
damage in DNA. By studying the electrochemistry of methylene blue, we gain insight into DNA-bound redox
reactions and DNA-mediated electron transfer processes. This process can be used as an electrochemical assay for
single base-pair mismatch detection and DNA sensing.
Live, Interactive Gaze and Speech: A Novel Paradigm to Investigate Social Cognition in Autism
Heesu Park
Mentors: Shinsuke Shimojo and Connie Wang
People with autism spectrum disorder (ASD) display atypical social behaviors in gaze and speech compared to
neurotypical (NT) controls. In this study, we investigated gaze and speech in live, real-time, face-to-face
interactions between pairs of subjects conversing freely as we recorded videos and monitored eye movements.
Automated gaze analysis requires detection of face, eyes, and mouth regions of interest (ROIs) in videos. Using
image-processing and feature-tracking techniques based on Kanade-Lucas-Tomasi (KLT) algorithm, I created code
to improve ROI detection. This raised detection accuracy to better account for natural range of head and body
movements during conversations. Previously, we found that gaze patterns differed between ASD and NT groups,
with reduced gaze to eyes in ASD (p<10-12). Using detailed ROI analyses, I found that ASD participants lacked two
prominent behaviors of NTs: 1) back-and-forth gaze between left and right eyes, and 2) shifts of gaze from
background into eyes, suggesting reduced social motivation to look at eyes. Beyond gaze, I also developed a plan
for automated acoustic analysis of speech, including quantification of fundamental frequency (F0), perceptual pitch,
pitch variation, and variability in voice characteristics over time. Speech, gaze, and their interplay will lead to
greater understanding of real-world social behaviors.
Live Imaging of Shadow Enhancer Activity via snail and sog Expression Changes in Drosophila
melanogaster Embryos
Hong Joon Park
Mentors: Angela Stathopoulos and Jeremy Sandler
Snail and sog are key Dorsal-Ventral patterning genes in Drosophila melanogaster. Curiously, these genes contain
cis-regulatory modules (CRMs) that are not only seemingly redundant to their promoter-proximal CRMs, but are
also located far from the genes they act on. This project aims to study the contribution of snail and sogs seemingly
redundant CRMs through live imaging of embryos using MS2-MCP and PP7-PCP plasmid constructs. The MS2
construct has been successfully completed for the sog distal enhancer, though the PP7 construct is currently under
development. With more time, the other constructs can be made and injected into embryos for live imaging.
Designing Inducible Systems for Controlled lncRNA Knockdown in HEK293T and E14 Mouse Embryonic
Stem Cells
Vincent Park
Mentors: Mitchell Guttman and Amy Chow
Long intergenic non-coding RNA (lincRNA) have recently been found to be crucial in regulating gene expressions
and cell state decisions in pluripotent stem cells. Two systems of controlling specific protein expression, Tet-On
FKBP-Shield1 and AID-auxin systems, have been explored to improve CRISPRi, a method that regulates gene
expression at the transcriptional level. In the Tet-On FKBP-Shield1 system, transcription of the FKBP destabilization

63

domain-tagged protein of interest (POI) is activated in the presence of doxycycline. Shield1 is then added to bind
to FKBP to inactivate protein degradation of the POI. The AID-auxin system functions so that AID degron-tagged
POI degrades in the presence of auxin. Applying these systems to control when the dCas9-KRAB that represses
transcription of target genes in CRISPRi is expressed would enable CRISPRi knockdown of both essential and
nonessential genes. To test these systems, HEK293T cells expressing OSTIR1-GFP-AID were treated with different
types of auxin. GFP fluorescence in transfected HEK293T cells did not degrade after the addition of auxin. This may
be due to overexpression of GFP or inefficiency in delivering auxin to GFP-AID. In addition, E14 mES cells
expressing Tet-On-GFP-FKBP were treated with varying doses of doxycycline and Shield1. Addition of doxycycline
and Shield1 does induce GFP expression, suggesting promise in incorporating the Tet-on FKBP-Shield1 system into
CRISPRi. Time courses of GFP fluorescence in E14 mES cells expressing TRE-GFP-FKBP will be measured to
understand the kinetics of the system. Further optimizations will need to be performed before attempting CRISPRi
with our newly created dCas9-KRAB-FKBP construct.
Interplay Between Usp16 and Wnt Pathway in Bone Development in Down Syndrome
Preethi Periyakoil
Mentors: Michael Clarke, Maddalena Adorno, and Thomas Miller
Earlier work done by our laboratory suggests that the gene Usp16, which is located on chromosome 21, plays an
important role in hindering self-renewal in stem cells of Down syndrome patients; on the other hand, Wnt signaling
is known to be a crucial pathway in stem cell self-renewal. The current study was undertaken to understand the
relationship between Usp16 regulation and Wnt signaling in human mesenchymal stem cells.We established a
culture of primary prenatal mesenchymal stem cells from a Down syndrome patient. In two of these cultures,
Usp16 was downregulated using lentiviral vectors coding for two different small hairpins for Usp16. In another
culture, the p16 gene, which is a downstream target of Usp16 also important for self-renewal, was downregulated
The last culture was infected with a control lentivirus. The infected cells were isolated by fluorescence-activated cell
sorting (because each lentiviral vector co-expressed the GFP gene), and they were treated with and without Wnt3a
at a concentration of XXX. The activity of the Wnt pathway was verified using qPCR for transcriptional targets of
Wnt, such as AXIN2, LGR5, and DKK2. The results of the qPCR were verified using TOPFlash, which is a reporter for
beta-catenin (a protein that accumulates in the nucleus when the Wnt pathway is active). Western blots were then
performed to detect beta-catenin and Usp16 in Down syndrome versus normal fibroblast cells, in the presence and
absence of Wnt. We learned that when Wnt was present, beta-catenin levels increased significantly in normal cells,
but they did not change in Down syndrome cells. This data suggest that an extra copy of chromosome 21 reduces
Wnt activity, and thus that Usp16 negatively regulates the Wnt signaling pathway, thereby preventing stem cell
self-renewal.
Void Collapse Near a Material Boundary in Gelatinous Substances Under Dynamic Loading
Anurag Pillutla
Mentors: Joanna M. Austin and Galina Shpuntova
Shock induced void collapse has a wide range of applications ranging from biomedicine to explosives. However, the
possibility of wave attenuation in a heterogeneous material prompts the need for studying void collapse under
stress wave loading. We focus on investigating void collapse under elastic-plastic stress wave loading in a model
two-dimensional setup using two tissue-surrogate polymer materials of varying acoustic properties. Series of
experiments are carried out to understand the hydrodynamic processes of void collapse near a downstream
boundary. Single-frame, two-colour particle image velocimetry and shadowgraph imaging techniques are used to
study parameters like velocity, vorticity, evolution of void shape and size in order to quantify differences between
the void collapse process in the presence and absence of a boundary.
Unambiguous Measurement of Linear Material Properties Using Nonlinear Responses of
Nanoelectromechanical Systems
Alex Place
Mentors: Michael Roukes and Peter Hung
The nonlinear dynamics of nanoelectromechanical doubly-clamped beams have often been measured to determine
nonlinear properties of nanomaterials. In this experiment, we demonstrate the use of nonlinear properties of
nanomaterials to measure the speed of sound, a linear nanomaterial property. Currently the speed of sound in a
nanomaterial cannot be determined independently of stress by measuring the fundamental mode of a
nanoelectromechanical doubly-clamped beam. While higher modes can be measured to determine the speed of
sound, measurements of the fundamental mode are more sensitive than measurements of any other mode. A new
paradigm to unambiguously determine the longitudinal speed of sound in a material, decoupling the effects of
stress while using only the fundamental mode, was developed in this experiment. In this experiment, the nonlinear
resonant frequency of doubly-clamped silicon beams was measured as a function of the displacement amplitude of
the beams. The speed of sound, and consequently, the Youngs modulus of each doubly-clamped beam was then
calculated. Geometrically identical doubly-clamped beams with various speeds of sound were fabricated by

64

changing the beams crystal orientations. The measured speeds of sound match the theoretical prediction of the
crystal orientation-dependent speeds of sound of silicon. These results verify this new paradigm to find the speed
of sound in nanomaterials and motivate further research in the nonlinear dynamics of nanoelectromechanical
resonators.
Nucleus Localization of a DNA-Binding Polyamide in Human Cancer Cells
Anand Poozhikunnel
Mentors: Peter B. Dervan, Tiezheng Jia, and Amanda Silberstein
Pyrrole-imidazole polyamides have been shown to be able to selectively bind to particular DNA sequences by
varying the arrangement of pyrrole and imidazole monomers. This binding interaction can cause a conformational
change in the DNA double helix so that it inhibits transcription and limits cell growth. This observation has led to
the exploration of the use of polyamides as tumor suppressors. In order for these polyamides to be effective,
however, they need to be able to reach the nucleus where DNA is stored inside a cell. In this experiment, the
polyamide will be synthesized and used to treat three different cancer cell lines-liver, endometrial and breast
cancer. Confocal microscopy techniques are used to determine the extent of nuclear localization.
Precision Measurements of Snow Crystal Growth Parameters as Inputs for Computer Modelling of Snow
Crystal Growth
Ryan Potter
Mentor: Kenneth G. Libbrecht
Full scientific understanding of snowflake growth is yet to be realised, as the many-body dynamics controlling the
growth of a crystal from the vapour phase are particularly complex. Here we report our progress on developing a
comprehensive computational model capable of simulating ice crystal growth in well-defined conditions based on
the use of cellular automata. The growth rates for the basal and prism facets of a snowflake grown at -10oC were
measured. The samples were grown from the initial conditions of the tips of thin needles which produced blockshaped snowflakes, and small thin crystals which produced plate-shaped snowflakes. Both of these types of sample
experienced similar growth conditions. Using the computational model of snowflake growth we then ran simulations
with input parameters to match the observed growth rates and morphologies for both data sets. We aim to run
models with similar input parameters but differing initial conditions, and hope that both observed morphologies will
be produced. This would provide evidence for the surface dependant attachment kinetics hypothesis. We
quantitatively compare experimental results with those from the computer model in order to improve our
understanding of the mechanics of snowflake growth and to make modifications to the existing model.
Modifying and Characterizing a Polyalphaolefin Flow Loop Intended to Cool Electronic Chips for Aircraft
Juliane Preimesberger
Mentors: Beverley McKeon and Tess Saxton-Fox
This project focused on altering an existing experimental setup designed to measure the convective heat transfer
coefficient of a prototype cooling plate in order to determine if the cooling plate could be used to cool electronic
chips for aircraft. A more efficient system to cool electronic chips is needed because of increased heat generation
from modern, complex circuit boards. Lockheed Martin proposed a prototype cooling plate made of offset metal fins
sandwiched between two aluminum plates cooled by polyalphaolefin (PAO). Previous work involved designing and
building a flow loop to test the cooling plate. This projects goals were to measure the pressure and temperature
change across the test section, thermally isolate the system from the environment, and design and implement a
cooling system that would keep the PAO at a constant temperature; the existing flow loop was modified to reach
these goals. Once these properties were quantifiable, experiments could be run to characterize the system, by
calculating the convective heat transfer coefficient of the prototype cooling plate.
Building an Apparatus for Analyzing Josephson Junctions for Superfluid 4He
Anjali Premkumar
Mentor: Keith C. Schwab
Superfluid 4He, or heliums quantum ground state attainable below 2.172K, produces Josephson oscillations, or
mass currents which are dependent on the quantum phase difference between two weakly linked bulk regions,
through appropriate Josephson junction materials. We have made progess in designing, building, and testing an
apparatus which will sensitively measure the Josephson current through various junction materials. Our apparatus
involves the use of a 4.2 GHz niobium resonator to measure Josephson current; currently, our resonator exhibits a
quality factor Q of about 3000 at 4K. Our resonator will thus have a sensitivity for superfluid 4He mass current on
the order of 10-14 kg, which is larger than that of previous detectors by a factor of ~ 1000. We aim to find a
junction material that allows for a large Josephson current at low temperatures. High-current Josephson junctions
can be used in superfluid helium quantum interference devices (SHeQUIDs) to create ultra-sensitive gyroscopes.

65

Direct Tests of a Pixelated Microchannel Plate as the Active Element of a Shower Maximum Detector
Federico Presutti
Mentors: Maria Spiropulu and Artur Apresyan
One possibility to make a fast and radiation resistant shower maximum detector is to use a secondary emitter as
an active element. We further the study of microchannel plate photomultipliers (MCPs) as an active element of a
shower-maximum detector. We present test beam results obtained using Photonis XP85011 and Photek 240 MCPs
to detect secondary particles of an electromagnetic shower. We focus on the use of the multiple pixels on the
Photonis MCP in order to find a transverse two-dimensional shower maxima distribution. A spatial resolution of 0.8
mm was obtained with an 8 GeV electron beam. A method for measuring time resolution from the Photonis MCP as
a whole is presented, and it improves the time resolution for individual pixel readouts. The time resolution is found
to be better than 40 ps.
The Dynamical Evolution of the GJ 876 Planetary System
AbhijitKrishna V. Puranam
Mentor: Konstantin Batygin
Gliese 876, also known as GJ 876, is a red dwarf star 15 light-years from Earth and is the closest known star to
harbor a multi-planetary system. This system provides an interesting case of extra-solar planetary dynamics, due
to the chaotic-Laplace resonance of the outer three planets and the high eccentricity observed in the inner planet.
This work used analytical methods and numerical simulations to characterize the short and long term evolution of
the GJ 876 planetary system. First, using the methods of secular perturbation theory, we developed an analytical
understanding of the phenomenon observed in simulations conducted using the Symba gravitational dynamics
software package. The predictions of the theoretical model were then corroborated through numerical simulations
using a version of Symba that was modified to consider relevant tidal effects. These results state the inner planet,
GJ 876d, cannot have the high eccentricity the observations indicate given the age of the system.
Glass-Ceramic Scintillators for Gamma Ray Detection Applications
Kira Pyronneau
Mentors: Brent Wagner, Zhitao Kang, and William L. Johnson
Scintillators, materials that emit photons in the presence of radiation, have proven to be useful in gamma radiation
detection technology. Currently, this technology is created using fragile and expensive single crystal scintillators.
However, glass-ceramic scintillators have shown potential for being a cheaper and more durable alternative to what
is currently available. For this project, glass-ceramic scintillators with various levels of the dopant samarium were
formed. Data was collected on the photon emission of the scintillators in the presence of radiation of various
wavelengths. Very defined emission peaks were observed at certain wavelengths in the spectrum. Also, as the
amount of dopant was increased in the glass samples, an increase in photon emission was observed. Defined peaks
with high amounts of photon emission are ideal for radiation detection in a device which indicates that glassceramic scintillators could be used for this application.
ClassLess: A Comprehensive Database of Young Stellar Objects
David E. Qu
Mentors: Lynne Hillenbrand and Nairn Baliber
Young Stellar Objects (YSOs) are stars in the early stages of evolution. This classification includes both protostars,
contracting fragments of molecular clouds that are still accumulating mass, and pre-main-sequence stars, which
have completed accretion but havent started burning hydrogen. Currently, there is no predictive model of star
formation describing the transition from protostar to the main sequence. Modern advancements in optical and
infrared spectroscopy as well as photometry at x-ray to millimeter wavelengths have produced a profusion of
observations. However, no convention for compiling the expanding literature of YSO data has been established,
unnecessarily scattering efforts to understand the phenomena. ClassLess is a web-based relational database that
aims to solve this problem.
Designed with an HTML5/Bootstrap/jQuery user-interface and powered by a Python/Django/PostgreSQL backend,
ClassLess allows YSO queries by any combination of coordinates, data columns, or numerical restrictions.
Furthermore, we provide plotting, data export, and external cross-references to sites such as Simbad and IRSA.
Our upload scripts give us the flexibility to dynamically determine the most fiducial measurements, so our queries
can represent the pinnacle of YSO data, condensed in one central repository.

66

1D Quantum Cellular Automata Using the 'Margolus Neighborhood' Update Method

Patrick J. Rall
Mentors: John Preskill, Ning Bao, and Nicole Yunger Halpern
Quantum physics often benefits from developing quantum versions of ideas in classical game theory. Cellular
automata are 1-player games commonly used in classical studies to generate complexity, so quantum cellular
automata can be similarly used to study the complexity of quantum many-body systems. The `Margolus
Neighborhood' cellular automaton update method has some advantages over other approaches like elementary
cellular automata because its classical limit is reversible, hence allowing a direct analogy between a closed
quantum system and a cellular automaton. Furthermore, a wide range of rules can be generated based on wellstudied quantum gates, hence allowing the cellular automata to serve as a testbed for testing measures of
entanglement complexity in a variety of settings. This analysis investigates complex network measures on graphs
formed by measures on two-qubit subsystems, assessing what can be learned from quantities like mutual
information and negativity. Additionally, the range of possibilities of the rule set is studied in terms of area laws
and correlation decay length.
Geometric Flows in Quantum Information: Ricci Flow in a Matrix Geometry
Preston Rasmussen
Mentor: Matilde Marcolli
We explore the Ricci flow in a finite-dimensional representation of a noncommutative torus, which has an
interpretation as a matrix geometry. We show that two formulations of the Ricci flow in this setting are equivalent
and use this fact to prove several results about the entropy of the system. We further study the properties of the
scalar curvature and the interpretation of these geometric properties in the setting of Quantum Information. In
particular, we explore the relationship between the Ricci flow and the relative entropy and how these are related to
both the flows of the quantum Fisher information metric and to the heat equation of the matrix geometry. It is
known that the time of the Ricci flow is not the physical linear time, but it is still not known whether or not the Ricci
time is exponential in nature.
Estimating Parameters of the TX-TL System to Model a Genetic Bistable Switch
Anushka Rau
Mentors: Richard M. Murray and Clarmyra Hayes
Biological circuits are networks of molecular and cellular components which can produce a specified protein or
macromolecule. They were first developed in vivo in E. Coli, but this process is inefficient, so in vitro systems were
developed. This project attempts to prototype these in vitro circuits (in a specific cell-free system called TX-TL)
using MATLAB to estimate parameters associated with reaction rates, focusing on the Tet system. Characterization
experiments of the Tet promoter, repressor, and inducer (aTc) were run in TX-TL to generate data, then that data
was fit to the existing MATLAB TX-TL model to estimate factors such RBS binding dissociation constant. Next, the
parameters generated by this as well as those generated by research on the Lac operon will be used to predict the
behavior of a genetic bistable switch of pTet-LacI and pLac-TetR and experiments will be run to test the accuracy of
the prediction. Our preliminary parameter estimation shows a fairly consistent fit of the model to experimental
data.
Probing the Reversibility of Hydrogen Oxidation in a Variant Oxygen Tolerant [NiFe]-hydrogenase
Gerri Roberts
Mentors: Fraser Armstrong, Emily Brooke, and Jonas Peters
Hydrogen has potential to be a clean and renewable fuel source. Before implementation, an efficient and costeffective hydrogen production and oxidation catalyst is required. In nature, hydrogenase enzymes catalyze this
reaction efficiently using the inexpensive and abundant metals, nickel and iron. Hydrogenase-1 (Hyd-1) is a [NiFe]hydrogenase from Escherichia Coli. At neutral pH, this enzyme exhibits catalytic bias towards hydrogen oxidation.
Oxygen sensitive counterparts have this catalytic bias, but to a lesser degree. At neutral pH, they are able to
catalyze both hydrogen production and proton reduction.
Using crystallography, genetic engineering, and protein film voltammetry, an amino acid canopy containing
arginine, the likely proton donor involved in the reaction mechanism, has been identified. In order to confirm the
importance of this arginine residue, amino acids within the canopy have been changed by site-directed
mutagenesis.
Previous work has demonstrated that lowering the pH enables native Hyd-1 to catalyze proton reduction. The work
presented here focuses on the replacement of two aspartic acid residues within the canopy, resulting in three
variant enzymes. The ability of these variants to catalyze proton reduction has been studied. Preliminary results
indicate that these two aspartic acid residues are influential in the enzyme mechanism.

67

Solvent Screening Effect on Band Positions of Functionalized Si(111) Surfaces

Connie Robinson
Mentors: William A. Goddard, Ravishankar Sundararaman, and Yuan Ping
Catalysis of water-splitting at semiconductor surfaces is an essential component in the development of a
photoelectrochemical cell which can perform artificial photosynthesis. A promising choice for such a surface is a
Si(111) surface functionalized with organic molecules. The dipole moment of the functional groups changes the
band edge positions of the surface, so the bands bend as they approach the surface. Since every surface site needs
to be terminated with a functional group in order to protect the surface, fairly electronegative groups must be used
to achieve the desired energy level, with small alkyl groups filling in the remaining sites. It is typically assumed
that the band edge positions are not affected by the environment, whether it be vacuum or mercury or aqueous
solution. Here we show that the dielectric response of the surroundings can result in a decreased effective dipole
moment of functional groups. Joint density functional theory (DFT) combines electronic DFT applied to the silicon
surface with another model applied to the solution, such as classical DFT or simpler models. We find that solvent
screening can shift the band positions by ~0.5 eV with a quarter of a monolayer of electronegative groups. We
predict mixed monolayers with ideal band positions in solution with solvent effects included.
Fabrication of Nanobeam Photonic Crystal Cavities in Nd:YVO for Efficient Light-Matter Interactions
Jake Rochman
Mentors: Andrei Faraon and Tian Zhong
Rare-earth ion doped crystals are of interest for quantum storage due to their long optical and spin coherence
times. However, these systems are characteristically slow and have weak light-matter interactions. Here, photonic
crystal cavities are fabricated in neodymium doped yttrium orthovanadate using focused ion beam milling to
increase the emission rates of the emitters and enhance the light-matter interactions. High quality-factors and
small mode volumes are realized within this fabrication scheme. A lifetime reduction from the Purcell effect is
measured. Cavity-enhanced light-matter interactions are observed as the cavity is tuned on resonance. These
results are promising for on-chip quantum memory or single ion qubits using rare-earth ions.
De Novo Genome and Transcriptome Assembly of Caulerpa prolifera
Roel Rodriguez
Mentors: Elliot Meyerowitz and Tyler Gibson
The genus Caulerpa, with its coenocytic cell structure and well-differentiated morphologies, provides a unique, yet
useful experimental system for the study of cell theory and growth kinetics. Caulerpa prolifera, the most studied
species, shows promise as a model coenocytic organism due to its large size. Internally, C. prolifera shows an
absence of septation. This raises the question of how Caulerpa prolifera can develop such an elaborate morphology
in the absence of cell-cell signaling and in direct defiance of modern cell theory. We utilize the de novo assemblers
Velvet and Trinity in order to assemble the genome and transcriptome (respectively) of C. prolifera for the first
time. Sequencing the genome of C. prolifera would allow us to address the structural anomalies of its unicellular
structure, and provide the genome for further research for others interested in Caulerpa.
Earth-Abundant Metal Catalysis as a New Paradigm in Chemical Synthesis
Andrew M. Romine
Mentors: Robert H. Grubbs and Anton A. Toutov
Manufacturing commodity and fine chemicals critically impacts both industry and academia. To avoid current
costly, wasteful, and energy-intensive methods, development of efficient, sustainable, and fundamentally new
approaches to the production of valuable chemicals relevant to multi-billion dollar industries such as
pharmaceuticals, agrochemicals, and materials needs to be addressed.
In early 2015, Toutov, et al., described a new chemical process combining simple, inexpensive bulk chemicals with
common potassium salts as the catalysts for producing industrially valuable silicon-containing substances (DOI:
10.1038/nature14126). Continued work in this field has accelerated since this publication and spans multiple
sectors: the production of next-generation radiopharmaceutical ([18F]-PET)probes, the discovery of powerful
sodium salt catalysts enabling the development of novel nitrogen-silicon and oxygen-silicon bond formation
protocols.
Recent studies show positive results for formation of oxygen-silicon bonds in alcohols, such as benzyl alcohol, and
nitrogen-silicon bonds in amines, such as N-methylbenzylamine. Work in PET probe development has enabled the
synthesis of eight unique radiopharmaceutical candidates, lending to future in vitro and in vivo studies.
Experiments with the milder potassium hydroxide also provide positive results, demonstrating occurrence of gentle
or even mono-selective silylation. Catalytic silylation remains a young field deserving attention due to its novel
solutions for academic and industrial problems.

68

Targeting First-Row Transition Metal Complexes Supported by a Catechol Diphosphine Ligand for Small
Molecule Reactivity
Jeffrey N. Rosenberg
Mentors: Theodor Agapie and Justin T. Henthorn
The terphenyl diphosphine catechol system (Cat-P2) previously developed within the Agapie group has shown small
molecule activity with transfer of multiple electrons and protons for Mo complexes. Previous attempts to access
first row transition metal complexes supported by this system have involved metalation of the Cat-P2 system with
Ni and Cu. The objective of this project is to establish organometallic systems on the Cat-P2 framework that will
show redox activity towards small molecules. Monometallic systems supported by the Cat-P2 framework can
undergo small molecule activation through transfer of protons and electrons from the central arene of the Cat-P2
system. This avenue has been pursued by developing Co complexes on the Cat-P2 framework in which the Co is
bound to the phosphines and the central arene ring. Bimetallic systems do not donate from the central arene but
instead activate small molecules by interaction with both metal centers. Cu and Ni monometallic systems explored
in the past in the Agapie group have both successfully been brought to bimetallic systems using Ru as the second
metal. Fe is also targeted as a source for a second metal.
Why Some Communities Are Susceptible to Radicalization
Helena Roy
Mentor: Erik Snowberg
I offer an interpretation as to why certain communities are more susceptible to radicalization than others. I
presume that vulnerable minority groups are unable to fully take part in the political process that governs them.
This offers the opportunity for a broker to represent the minority group. Ultimately, the broker benefits from
radicalizing a fraction of the minority cementing his position due to direct and indirect effects of distancing the
minority from its government. Pushing radicalization too far, however, can put the broker out of power due to
reactions from within the minority group and government. This is observed in the political behavior of some
religious sects and ethical groups.
Synthesis and Characterization of CyanthiwiginGagunin Hybrids
Yuka Sakazaki
Mentors: Brian Stoltz and Kelly Kim
The cyanthiwigin diterpenoids are natural products that have been isolated from the sponges Epipolasis reiswigi
and Mermekioderma styx, and there are currently more than 30 of them known. Almost all cyanthiwigins share a
fused 5-6-7 tricyclic carbon structure, and many of them display cytotoxicity and other biological activities. The
gagunins are natural products that share a similar fused 5-6-7 tricyclic structure, but with a more densely
oxygenated carbon skeleton than those of the cyanthiwigins. These oxidation patterns seem to heavily influence
biological activity, since gagunins display a wide range of cytotoxicity while sharing the same carbon skeleton.
These trends may be explored by testing biological activities of similar molecules with varying oxidation patterns.
Previously, our lab has developed a concise synthesis of cyanthiwigins B, F, and G through a common intermediate.
Here, we extend this route to the synthesis of cyanthiwigingagunin hybrid molecules by using conventional
oxidation techniques such as epoxidation and dihydroxylation. After characterization, the biological activities of
these cyanthiwigingagunin hybrids will be assessed in collaboration with various organizations.
The Stability of Quasicrystals With Soft Multiscale Interactions
Samuel Savitz
Mentors: Gil Refael and Mehrtash Babadi
Quasicrystals are exotic states of matter that are ordered but not periodic. Self-assembly of particles with soft
interactions into quasicrystalline structures is a relatively recent topic of much interest. While simple criteria have
been proposed for the stability of these quasicrystalline states, they rely on oversimplifying assumptions and are
incomplete. In this work, we revisit these criteria, critically assess their underlying assumptions and extend them in
several directions. Our revised criteria are supported by analytical and numerical evidence. Finally, we propose a
mechanism for realizing soft multiscale interactions in cold atom systems via simultaneous Raman dressing of
ground state atoms into two Rydberg states. Our numerical simulations of this system exhibit self-assembly of the
doubly-Rydberg-dressed atoms into a dodecagonal mosaic quasicrystal state only once our new criteria are met.
Quantum Entanglement With Trapped Ions
Drew Schffer
Mentors: Boris Blinov and David Hsieh
Quantum entanglement has been the target of a disproportionate amount of the metaphysical angst surrounding
quantum mechanics since Einstein, Podolsky, and Rosen (EPR) first identified it in the 1930s. They believed it to be
a clear sign that quantum mechanics is an incomplete theory, since it implied that the commonsense postulates
known as local realism. This was left unresolved until John Bell proved the Bell Inequality, which provided an

69

experimental test of whether quantum mechanics is a local theory or not. However, various loopholes are present
in all such experiments that have been conducted. By using electrodynamic traps to isolate barium ions we hope to
perform the first loophole-free test of the Bell Inequality. In this project, the ion traps were calibrated to eliminate
micromotion, a major impediment to collection of quality data from the traps. The micromotion was largely
elimination, and our work in the future is to begin data collection.
Computation and Construction in the Chemical Reaction Network-Controlled Tile Assembly Model
Nicholas Schiefer
Mentor: Erik Winfree
Tile-based self-assembly and chemical reaction networks provide two well-studied models of scalable DNA-based
computation. Although tile self-assembly provides a powerful framework for describing Turing-universal selfassembling systems, assembly logic in tile self-assembly is localized, so that only the nearby environment can
affect the process of self-assembly. We introduce a new model of tile-based self-assembly in which a well-mixed
chemical reaction network interacts with self-assembling tiles to exert non-local control on the self-assembly
process. Through simulation of multi-stack machines, we demonstrate that this new model is efficiently Turinguniversal, even when restricted to unbounded space in only one spatial dimension. Using a natural notion of
program complexity, we also show that this new model can produce many complex shapes with programs of lower
complexity. Most notably, we show that arbitrary connected shapes can be produced by a program with complexity
bounded by the Kolmogorov complexity of the shape, without the large scale factor that is required for the
analogous result in the abstract tile assembly model. These results suggest that controlled self-assembly provides
additional algorithmic power over tile-only self-assembly, and that non-local control enhances our ability to perform
computation and algorithmically self-assemble structures from small input programs.
Evaluation of the Optical Properties of NiOx Protective Coating Under Working Conditions Using
Operando Optical Spectroscopy
William Schmidt
Mentors: Nathan S. Lewis and Ke Sun
The process of artificial photosynthesis uses solar energy to split water into oxygen and hydrogen in acidic or basic
conditions. Traditional photovoltaic materials with small energy band gaps are unstable under such conditions and
thus require additional protective coating to increase durability for photoelectrochemical solar fuel conversion.
A nickel oxide coating has been shown to extend the life of the device from a few hours to over 1000 in 1.0 M KOH,
but its optical properties under the conditions are less explored.
In this SURF project, we are using Operando optical spectroscopy, particularly ellipsometry, to study the optical
properties of the protective film. This would allow us to directly reconstruct the optical parameters of the films
under water oxidation conditions, which will help to understand the degradation of protective coatings and design
stable photoelectrodes for solar fuel conversion.
Experiments on samples from different deposition conditions in pH 9.5 solution reveals that under bias near water
oxidation voltage the amplitude ratio of s and p waves increases at higher energy wavelengths, while the peak
phase difference decreases. This change is more pronounced on samples prepared at low temperature and high
deposition rate, while less significant on high temperature and slow deposition.
Characterizing the Regulatory Effect of the Iron-Sulfur Cluster in DNA Polymerase Delta
Joseph Schneider
Mentors: Jacqueline K. Barton and Phillip L. Bartels
Recently, iron-sulfur clusters have been discovered in DNA binding proteins with the unusual property that these
clusters are only redox active when their host protein is bound to DNA. One such protein is the eukaryotic B family
polymerase delta, which contains a [4Fe-4S] cluster that can only be oxidized when the polymerase is bound to
DNA. Preliminary studies on the function of this iron-sulfur cluster indicated that oxidation of the cluster may be a
regulatory element to turn down or shut off DNA replication during times of oxidative stress. To test this theory,
further work was done to characterize the activity of polymerase delta with both a reduced cluster and an oxidized
cluster. Oxidation of the natively reduced polymerase is achieved either by bulk electrolysis through DNA tethered
to a gold electrode or by tethering anthraquinone to DNA. Extension assays are then performed by incubating the
polymerase with primed DNA and dNTPs for various time points. Either the primer or the dNTPs are radiolabeled,
allowing imaging of the extension products via gel electrophoresis. This established that oxidation of the iron-sulfur
cluster does indeed depress DNA replication, and allowed for quantitative characterization of this effect.

70

Quantifying Ubiquitination of Histone H2B in Single Cells by Next-Generation Sequencing

Jinyoung Seo
Mentors: Long Cai and Chang Ho Sohn
In this project, we aim to develop a high-throughput in situ proteomic tool to quantify a protein of interest in single
cells. The key to this project is to (i) in situ ligate a single-stranded DNA barcode to a protein of interest via
genetically encoded SpyTag-SpyCatcher chemistry, (ii) tag cellular index sequences to the DNA barcode of each
cell, and (iii) amplify and count the absolute numbers of the target protein using next-generation sequencing-based
molecular counting. As a proof-of-concept demonstration, we will measure relative abundance of ubiquitination of
histone H2B in individual cells.
Regioselective Wacker Oxidation of Simple Meso Dienes to Form Enantioenriched Allylic Alcohols
Jacob Sertich
Mentors: Robert H. Grubbs and Crystal Chu
Regioselective Wacker oxidation of a meso diene bearing a protected secondary alcohol was utilized to produce
enantioenriched allylic alcohols. By employing Pd(OAc)2 and a chiral phosphoric acid, a chiral catalyst was
developed for use in the desymmetrization reaction. Optimization studies were focused on the catalyst loading of
the chiral phosphoric acid and variations in the solvent system. The chiral phosphoric acid is hypothesized to act
catalytically in association to cationic palladium, leading to enantioenrichment observed in the allylic alcohol
product. Future studies involve expanding the substrate scope to include a diverse array of secondary and tertiary
allylic acohols. This work has potential to transform simple and readily available achiral starting materials to
relatively complex chiral products useful in target-oriented synthesis.
Fruit Flies Activity Recognition Using Decision Tree
Ruoqi Shen
Mentor: Yisong Yue
One technical challenge neuroscientists and biologists face is the fact that analyzing behavior requires analyzing
large amount of video or tracking data. For example, currently there are thousands of hours of video on fruit fly
behavior here at Caltech, and it is impossible for scientists to manually analyze this video. Machine learning is very
promising because it can be used to develop an automated approach for animal behavior analysis. This project use
decision tree to automatically detect behaviors such as lunge, tussle and other aggressive actions and copulation
attempt, copulation and other courtship actions in fruit fly tracking data. I first use features from single frame to
train models and then use features from multiple frames. I also train sliding window models and compare them
with the frame-level prediction models.
Biological Imaging Using Genetically Encoded Magnetic Contrast Agents
Jiemin Sheng
Mentors: Mikhail G. Shapiro and Pradeep Ramesh
Though optical microscopy is a powerful modality of imaging, its applications are limited by the penetration depth
of light in tissue. To circumvent this limitation, we wish to exploit the robust, non-invasive nature of magnetic
resonance imaging (MRI) in order to observe interesting systems biology, in vivo. We are working to create
genetically encoded magnetic nanoparticles in order to enhance T2 (spin dephasing) contrast in MRI. This is done
by borrowing the genome of the magnetotactic bacteria (MTB), a bacteria that creates magnetic particles, called
magnetosomes, in order to help it align along the Earths magnetic field. By engineering, transplanting, and
targeting a minimal set of the MTB genome, along with other helper proteins, into mammalian cells, we hope to
create a robust contrast agent which will allow us to image tissues in vivo and investigate processes such as tumor
metastasis, which has been very hard to do thus far.
New Manganese and Rhenium Complexes for CO2 Reduction Catalysis
Thomas Sheridan
Mentors: Harry Barkus Gray and James Blakemore
Manganese and rhenium carbonyl complexes with bipyridine ligands have been shown in prior work to function as
electrocatalysts for CO2 reduction. In our new work, related compounds bearing the analogous nitrogen-rich
bipyrimidine (bpm) or bipyrazine (bpz) ligands were synthesized, characterized, and tested for their ability to
catalyze CO2 reduction. Our new compounds include [Mn(CO)3(bpm)Br], [Mn(CO)3Br(bpm)Mn(CO)3Br],
[Mn(CO)3Br(bpm)Re(CO)3Br], and [Mn(CO)3(bpz)Br]. These new compounds were characterized by 1H NMR
spectroscopy, mass spectrometry, and single-crystal X-ray diffraction. Cyclic voltammetry showed promising
catalytic activity for a number of complexes, with the most robust catalysis found with the Mn-bpm monomer and
Mn-bpm-Re dimer. Results from on-going experimental work will be presented, including characterization of the
products of CO2 reduction.

71

Fabrication of Metallic Microlattice With Surface Deposited Perovskite Catalysts as Cathodes in Li-O2
Batteries
Jingwei Shi
Mentors: Julia Rosolovsky Greer and Chen Xu
The ability to store electricity is a crucial component in the transition to sustainable energy. A promising battery
candidate is the lithium-air battery, which utilizes the reversible reaction of lithium and oxygen to form lithium
peroxide to release and store electricity. However, several limitations prevent the widespread implementation of
such batteries, in particular the degradation of solvent and other organic species in the battery. Lanthanum based
perovskites have been identified as good candidates for catalysts that reduce the amount of solvent degradation.
In this experiment, we demonstrate the feasibility of attaching said perovskites to a metal cathode, providing a
platform for testing catalysts with minimal use of unstable organics.
Computation of Heegaard Floer Correction Terms
Laura Shou
Mentor: Yi Ni
The Heegaard Floer correction terms is an invariant for rational homology spheres, introduced in 2003 by Ozvth
and Szab as part of Heegaard Floer homology. We develop a computer program to compute the Heegaard Floer
correction terms for two types of 3-manifolds: the double branched cover of an alternating link L in S^3 and a
Seifert fibered rational homology sphere specified by data {e; (p_1,q_1),...,(p_r,q_r)}. We generate a quadratic
form, which is then used to compute the correction terms as described by Ozvth and Szab. Progress has also
been made towards computing the correction terms of a manifold S^3_{p/q}(K) obtained by Dehn surgery with
slope p/q on a knot K in S^3. The method would use grid diagrams and the combinatorial description of knot Floer

homology in a paper by Manolescu, Ozsvth, and Sarkar to compute the homology of the chain complex CFK (S,
K), and then a paper by Ni and Wu to compute the correction terms.
Further Work on Locally Recoverable and Constrained Codes
Kevin Shu
Mentor: Babak Hassibi
Recently, there has been interest in extending classical coding theory in order to apply it to modern distributed
storage systems. Two models for this problem are locally recoverable codes and constrained codes. We present
additional work on these topics and possible ideas for later research, including new algorithms and constructions.
We develop a class of constrained codes based on trees that have good systematic constructions. We also develop
ideas for a new decoding scheme for Tamo-Barg codes.
Effective Stiffness Modeling of Periodic Lattice Meta-Materials
Ritwik Shukla
Mentors: Dennis M. Kochmann and Alex Zelhofer
Periodic lattice meta-materials allow a wide range of possibilities to explore and design for efficient dynamic and
static properties. This includes characteristics such as stiffness under the action of dynamic and static loading.
Commonly lattices are modeled as geometrically perfect whereas in practice are susceptible to geometric
imperfections resulting in significant material property degradation. In this work an in-house finite element code is
used to model 2D periodic lattices subjected to geometric imperfections in the form of stochastic lattice member
removal. Effects of unit cell design, lattice member cross-section, separation of unit cell from material length
scales, and percent of lattice members removed are studied. Results from this study provide insight into designing
imperfection tolerant lattice meta-materials.
Environmental Context Dependent Memory in Virtual Reality
Ayaana Patel Sikora
Mentors: David M. Krum and Mathieu Desbrun
As virtual reality technologies are increasingly being used in applications such as training and education, it is
becoming crucial to understand the perceptual and psychological effects of virtual learning. Previously conducted
research in human memory has identified the environmental context dependent memory effect, in which
environmental stimuli are incidentally processed during learning and can later cue recall and improve memory
performance. We are exploring how this effect plays a role in virtual learning. We have developed study protocols
involving virtual reality environments and word recall tests to examine this effect, and we are currently validating
our experimental infrastructure. We have also constructed virtual replicas of real world environments using a 3D
laser scanner. These replicas will be used to test if virtual reality could evoke the environmental contexts for
learning and later produce improved recall in the corresponding real world environments. We also plan additional

72

studies to examine whether real world sensory cues, such as olfactory and auditory signals, could bleed in through
the immersion and provide additional context. Understanding how human memory is affected by environmental
contexts and simulated contextual cues could be beneficial to addressing learning and memory issues in domains
such as eyewitness testimony, psychotherapy, and education.
Topological Order in Marginally Self-Correcting Quantum Memory
Karthik Siva
Mentors: John Preskill and Beni Yoshida
The primary challenge for constructing large-scale quantum computers is the need to protect qubits from
decoherence, the process by which the information encoded in the entanglement within the hardware is lost due to
coupling with the environment. As a result, the feasibility of self-correcting quantum memory, a hypothetical
quantum memory device that could store information for arbitrarily long periods of time without the need for
repeated, active error-correction, has been a subject of intense study. This problem also appears closely linked to
that of the stability of topological order above zero temperature. In this project, a class of marginally selfcorrecting quantum memory models are shown to lack topological order above zero temperature. In particular, the
proof technique employed considers error-correction in ground states of defective versions of the full Hamiltonian
describing the memory model.
The Gut Bacteria Lactobacillus brevis Influences Hyperactivity in Drosophila
Matthew D. Smalley
Mentors: Sarkis Mazmanian and Catherine Schretter
It is well known that the microbiome holds a great deal of influence over its host organism. These gut bacteria can
play important roles in metabolism, immunity, and social behavior. Axenic flies (with no gut bacteria) are
hyperactive compared to regular flies. Normal locomotor activity is only restored when the bacteria Lactobacillus
brevis is present in the fly gut. By measuring gene expression in brain and gut neurons and performing locomotion
assays we examine the mechanism by which L. brevis is able to induce this change in fly behavior.
Measuring the Atmospheric Ground Layer at Palomar Observatory
Tasman Smith
Mentor: Dan McKenna
In order to determine if High Point at Mount Palomar is a valid site for the development of a new telescope, it is
important to measure the expected seeing at the location. By measuring the difference in temperature between
two points very rapidly and with high precision, it is possible to infer the expected degradation in image quality due
to changes in air density as a function of height. An array of thermocouples pairs mounted onto a fire watch tower
in the area provides us with a platform to measure precision 24-bit samples at 80 times per second. Preliminary
trials have shown that there is a standard deviation of 0.005 C between measurements of the same temperature,
with fluctuations of temperature around 0.1 C across 1 meter. In the near future, deployment of 20 thermocouple
pairs on the tower will provide us with the data to conclude if high point is a viable location for future construction.
Resonating Valence Bond States of Majorana Zero Modes
Jun Ho Son
Mentors: Jason Alicea and Ryan Mishmash
Originally conceived by Anderson in 1973, the resonating valence bond (RVB) wavefunction has stood as the
prototypical example of a quantum spin liquid state ever since. In its simplest incarnation in the context of spin-1/2
quantum magnets, the RVB takes the form of an equal amplitude superposition of all nearest-neighbor dimer
coverings, where here a dimer represents placing the two spins into a maximally entangled spin singlet state. In
this work, we consider generalizations of the canonical RVB concept to systems in which the spin degrees of
freedom are replacednby non-Abelian Ising anyons binding Majorana zero modes. In our Majorana-RVB(M-RVB)
state, a dimer represents fusing two Ising anyons into a particular, common fusion channel, say the identity
channel (equivalently, the physical Dirac fermion formed by the two Majorana zero modes is taken to be, say,
unoccupied). We proved that the M-RVB wave function on the 2D planar (open boundary condition) square lattice
is suitable for Monte Carlo calculations, i.e., it is sign problem free. Based on this construction, we present largescale numerical calculations of the M-RVB state and show that two-point Majorana-Majorana correlations decay
exponentially with separation distance, while dimer-dimer (i.e., parity-parity)correlations decay as a power law.
While this indicates that the square lattice state would be the ground state of a gapless local Hamiltonian, we
postulate that placing the problem on a frustrated lattice may lead to a fully gapped, nontrivial topological phase.

73

Investigating the Role of Innexin Genes in Supporting CVM Cell Migration in Drosophila
Ge Song
Mentors: Angelike Stathopoulos and Frank Macabenta
Collective cell migration is a very important process during embryo development, with directional movement
regulated by spatial and temporal expression of genes. In the fruit fly Drosophila melanogaster, the longest type of
cell migration is accomplished by caudal visceral mesoderm (CVM) cells, which originate from the posterior-most
end of the embryo and migrate synchronously toward the anterior as two distinct groups on either side of the body.
Little was known about the interactions between cells inside this migrating collective -- to gain insight, we
investigated the role of gap junctions, composed of a family of proteins known as the innexins (Inx) in flies. We
applied RNAi using GAL4-UAS system to knockdown each of the 8 different innexin genes specifically in CVM cells
and used antibody staining to observe the phenotype of CVM cell migration in embryos. Preliminary data showed
that some mutants of these innexin genes lead to slight dispersed or asymmetric phenotypes of CVM cell
migration, indicating that gap junctions may be one of the factors that keep the cells moving congruously inside
the migrating group.
A Cryogenic Blackbody Calibrator for Measuring the Optical Efficiency of Antenna-Coupled Bolometers
Rita Sonka
Mentors: Jamie Bock and Bryan Steinbach
The successful detection and characterization of the B-modes in the Cosmic Microwave Background (CMB) would
dramatically illuminate the physics of the inflationary era. The Observational Cosmology Group is iterating on
bolometers in an attempt to detect this signal. A cryogenic blackbody calibrator was created to determine the
optical efficiency for each new set of detectors with an internal load. Since the detected radiation does not go
through the optics in an internal-load, this separates the efficiency of the detectors and the optical setup. Two tiles
of detectors were then tested; Tile 1 has likely = 0.814 0.995, and Tile 2 has likely = 0.496 0.694.
Expression Profile Map of Dprs and DIPs in the Drosophila Larval Nervous System
Ekin Sonmez
Mentors: Kai Zinn and Robert Carrillo
The defectve proboscs response (Dpr) and Dpr-nteractng proten (DIP) subfamilies belong to the immunoglobulin
superfamily (IgSF) of cell surface proteins. Members of the IgSF are required for cell migration, axon guidance,
synaptogenesis, and synaptic plasticity. A high-throughput screen examined pairwise interactions among the
Drosophila Ig-domain proteins and identified the Dprs and DIPs as binding partners. dpr1 was previously shown to
be required for a behavioral response to salt aversion and for male courtship. The Zinn lab has also begun to study
other binding partners. However, not much is known about the expression and localization of the 21 Dprs and 9
DIPs. The aim of this project is to determine the cells expressing the Dprs and DIPs in the nervous system and
their specific localization within those cells. We will utilize Dpr- and DIP-GFP fusions and immunohistochemistry to
create an expression profile in the larval nervous system. At the end of this study, we hope to know the Dprs and
DIPs that are expressed in the nervous system and possibly identify interacting cells that express corresponding
Dpr and DIP binding partners.
Numerical Solver for the Linear Hamilton-Jacobi-Bellman Equation
Ulf Elis Stefansson
Mentors: John Doyle and Yoke Peng Leong
The Hamilton-Jacobi-Bellman equation provides globally optimal solutions to a wide range of control problems.
However, algorithms obtaining such solutions typically exhibit the curse of dimensionality enabling calculations for
just a moderate number of states. This paper presents a toolbox for solving a linear version of the Hamilton-JacobiBellman equation. The curse of dimensionality is addressed using tensor decompositions, where the resulting
algorithm scales linearly with the number of states in the system. Alternating least squares, the main algorithm in
the toolbox, is investigated and improved. Examples illustrate the performance of the toolbox.
Adapting an Optogenetic Tool for Large Scale Neuronal Activity Imaging With Single-Cell and
Millisecond Resolution in Subcortical Structures
Gregory S. Stevens
Mentors: Viviana Gradinaru and Nicholas Flytzanis
The field of neuroscience lacks the critical ability to simultaneously measure the activity of thousands or millions of
individual neurons with the temporal resolution of milliseconds in mammalian subcortical structures.
A recently developed fusion protein, CaMPARI (calcium-modulated photoactivatable ratiometric integrator), was
designed such that both free intracellular calcium and experimenter-controlled violet light were necessary for
permanent photoconversion. CaMPARI was then used to image the activity of a large population of cortical neurons
through a cranial window in mice.

74

The next breakthrough is to combine CaMPARI with fixing and clearing techniques, specifically PACT (passive clarity
technique), in order to enable large scale subcortical activity readout. This project focused on optimizing a variety
of protein expression, photoconversion, and fixation parameters to enable post-fixed imaging of CaMPARI. We
present results from mouse experiments that consisted of in vivo optogenetic activation of excitatory projections to
the VTA and post-fixed imaging of CaMPARI expressed in the VTA. Future work includes implementing behavioral
paradigms to discern differences in reward structure activity during different aspects of a reward (anticipatory,
consummatory, etc.) These initial experiments will hopefully be the basis for future widespread use of CaMPARI in
all brain structures.
Simulation Pipeline for Velocity Field Measurements Using the Kinetic SunyaevZeldovich Effect
Yubo Su
Mentor: Sunil R. Golwala
Various competing cosmological models are distinguishable only through precision measurements because their
differences are small; however, understanding their correctness is pivotal for understanding phenomena such as
the accelerating expansion of the universe. Measurements using the kinetic Sunyaev-Zeldovich (kSZ) effect are
expected to provide such precision with different systematic sources of error than current techniques. However, the
obstacles facing kSZ detection are currently not well understood. A full simulation pipeline is essential to
characterize these difficulties. This study focuses on subtraction of point source contamination in telescope images
using a realistic galaxy distribution from prior studies and the subsequent detection of various SZ effects in the
cleaned image. To improve robustness of the subtraction process, images from multiple frequency ranges are
cross-correlated to find sources using all available frequencies. Varying emissivities of sources is also accounted
for.
Epoxidation Reactions With a New Titanosilicate, Molecular Sieve Catalyst
Alp Mehmet Sunol
Mentors: Mark E. Davis and Joel Schmidt
Synthetic molecular sieves are microporous materials that are widely used as catalysts. The Davis group has
synthesized a new titanosilicate, Lewis acid molecular sieve catalyst, denoted Ti-STW. The STW framework has
chiral, helical channels, but in bulk quantities only exists as a racemic mixture. The group is currently developing a
method to synthesize bulk quantities of an enantiopure STW catalyst. Enantiopure STW has the potential to impart
its chiral nature in reactions, leading to enantioenriched products. Titanosilicate molecular sieves are known to be
catalytically active for epoxidation reactions, which lead to chiral products. Racemic Ti-STW was tested as an
epoxidation catalyst to determine: 1) if Ti-STW shows catalytic activity for epoxidation reactions, 2) what size of
reactants fit into the pores of STW and 3) how to optimized reaction conditions. Various olefin-containing
substrates were reacted with a range of oxidants, using Ti-STW as the catalyst. Reactions were monitored by 1H
NMR, and conditions were optimized such that three allylic alcohols trans-2-hexen-1-ol, 3-methyl-2-buten-1-ol,
and allyl alcohol reacted to form epoxides with significant yields. The optimized reaction conditions developed
with three different starting reagents will be used to test for enantioselectivity when enantiopure STW is prepared
in the future.
A Computational Simulation and Analysis of Spinal Cord Stimulation Patterns
Atli Thor Sveinbjarnarson
Mentors: Joel W. Burdick and Ellen Feldman
Spinal cord stimulation has been shown to be a viable treatment for paraplegics with spinal cord injury. The
stimulation is achieved through electrical pulses delivered via electrode arrays which are either placed on the
patients back or implanted in the epidural space over the lumbosacral spinal cord. The polarity and voltage of
individual electrodes in an array can be controlled, forming different stimulation patterns. The patients
performance is highly dependent on the selected stimulation pattern. In order to better understand how different
stimuli give rise to different performance we created a computational model of the stimulation process. A 3D
volume conductor model accounting for the geometry and electrical properties of the spinal structure is created.
For each stimulation pattern the electrical potential within the model is calculated by solving Laplaces equation
numerically with the appropriate boundary conditions imposed by the electrodes. The electrical potential is used
with a model of the nerve fibers to calculate the activation of spinal cord regions and individual spinal roots. The
results of the simulations could be used with clinical data to identify important stimulation regions which might lead
to improved array designs and a more effective selection of stimulation patterns.

75

Using Electromagnetic Integral Equation Solvers With Optimization Algorithms to Design 3D Antennas
Amal Tariq
Mentors: Ali Hajimiri and Constantine Sideris
The development of a technique to effectively design antennas that meet or exceed target constraints would be
extremely useful to the production of modern day technology. Optimal antenna design is hindered by the fact that
there are no analytical solutions to Maxwells equations for anything beyond the simplest of problems. However,
using the power of modern computing, complicated problems can be solved through discretization. In order to
design PCB or IC antennas that meet specific criteria, an integral equation solver for 3D dielectric surfaces was
developed and will be used in conjunction with an optimization algorithm. The integral equation solver uses a
triangular patch model to discretize arbitrarily shaped 3D surfaces and applies the RWG basis functions to generate
a method of moments solution to the EFIE and MFIE. An iterative optimization algorithm will be applied with the
integral equation solver to design antennas to meet specific requirements. Iterative algorithms work well for such
problems by improving upon previous structures until further changes do not provide appreciable increases in
performance.
Characterization of Cis-Regulatory Modules in Sea Urchin Skeletogenic Effector Genes
Kisha Thayapran
Mentors: Eric Davidson and Lijun Wang
Gene regulatory networks are composed of genes and the regulatory inputs that interact to direct the expression of
those genes. Cis-regulatory modules (CRMs) direct expression of the genes they are associated with by dictating
how, when, and what inputs will affect expression. In this study, we attempt to characterize the CRMs of 11
skeletogenic effector genes in Strongylocentrotus purpuratus. Regions of intronic and intergenic sequences found
to show significant levels of activity in previously done Nanostring experiments in each gene were deemed hot
pieces. In a deletion assay, regions of each hot piece containing transcription factor binding sites were deleted,
and each construct was injected into embryos to analyze CRM activity. In a mutation assay, transcription factor
binding sites were mutated on each hot piece. These mutated constructs will be injected into embryos as part of
Nanostring experiments to further characterize CRM activity. Furthermore, the results of these two assays should
provide insight into the role CRMs play in spatial and temporal expression, as well as the level of expression of
each gene.
Detection and Analysis of Fast Radio Burst Gamma-Ray Emission
Rachel Thorp
Mentors: Pall Jakobsson and Gunnlaugur Bjornsson
Fast radio bursts (FRBs) are radio transient emissions that have only been recently discovered. As such, very little
is known about their nature. We use Fermi-LAT data for eight recently detected FRBs to perform both the likelihood
analysis of various source models, as well as the photon flux emission upper limit calculations for these events. The
sources are likely of celestial origin with redshift values ranging from 0.5 to 1, and have been found to only last for
approximately four milliseconds. The values obtained do not currently support a particular source model for any of
the eight sources, although we may be able to use these values to eliminate some hypothesized emission models.
By having a better understanding of the FRB source structures, we hope to aid in the characterization of the
baryonic content of the universe.
A Computational Evaluation of the Whorfian Hypothesis in Spatial Relations
Christine Tseng
Mentors: Terry Regier, Yang Xu, and Markus Meister
When does the structure of a language influence its speakers' perceptions of the world? Traditional approaches to
the Whorfian hypothesis have focused on assessing the presence of language-specific effects but left the condition
under which they are most prominent undetermined. We hypothesize that the degree of the Whorfian effect
depends on perceived structures in the world; specifically, that language exerts more influence when world
structures appear ambiguous. We test this hypothesis in the domain of spatial relations. Using computational
analyses at group and individual levels, we show that the structure of linguistic categories plays a greater role in
predicting similarity judgements of ambiguous spatial relations. Our study therefore provides a quantitative
assessment of and support to the "name strategy" as an unconscious cognitive mechanism underlying the Whorfian
effect.

76

Investigating the Thermal History of Australasian Tektites by Studying Lechatelierite Inclusions

Richard Turley
Mentors: Edward Stolper and Catherine Macris
Tektites are impact glasses formed during complex processes of melting and chemical mixing in the vapour plume
produced by early crater excavation following the collision of a large, hypervelocity projectile with the Earths
surface. Oblique collision of such a projectile causes geographical dispersal of tektites to form strewn fields. This
project examined tektites from various localities in the Australasian strewn field to study a hypothesised
relationship between their thermal history and spatial distribution. The focus was on investigation of chemical
diffusion between lechatelierite inclusions (pure SiO2 glass generated by the shock melting of quartz) and the
tektite host glass (of approximately rhyolitic composition). Using an electron probe micro-analyser (EPMA), majorelement profiles were created of transects extending from the lechatelierite core into the host matrix. Profiles show
smooth curves with a rapid decrease in SiO2 concentration at the lechatelierite/host contact, indicative of diffusion
between the lechatelierite and host glass. These profiles will be compared to an existing experimental dataset of
profiles from synthetic tektites, created at known temperature/time conditions in an aerodynamic levitation laser
furnace. This will allow estimation of the thermal conditions of the tektites formation and give unique insights into
plume dynamics.
Finding Accurate Variance Models for Turbulent Premixed Flames
Tomas Tussie
Mentors: Guillaume Blanquart and Simon Lapointe
Combustion is responsible for 79% of the worlds energy production. DNS of an entire combustion engine is highly
expensive, and LES, although cheaper, relies on a low-pass filter to provide results. Important non-linear qualities
of a flame cannot be determined without knowing the variance of the temperature. It is known that for passive flow
1
the constant Cs is in the range of to 0.25, so it is important to know whether the same holds true for reacting
12

flows. Using data from DNS simulations, various box filters with different filter widths were used in order to
calculate both the modeled variance and the exact variance of the flame. Linear regressions were run on these
values in order to predict the values of Cs for different filter widths. Using this data and characteristics from the
flame, it was found that the values for different filter widths collapse onto a single curve, suggesting that a function
could be found to use in algebraic models.
Domain Adaptation for Classification in Astronomy
Samarth Vaijanapurkar
Mentors: Ashish Mahabal and Adam Miller

The data from the public surveys and the light curves generated for the objects observed are different for different
surveys based upon their filters, brightness sensitivities and such other criteria. In light of many new surveys which
are still pending for classification and many more upcoming surveys (LSST, ZTF etc.) which are much more
detailed and fainter than the existing surveys, so a generalized method is needed for classification of variable stars.
Herein this project we are exploring the application of Domain Adaptation for classification problem. The
classification model works with a good performance when operated over a single survey, where the data is ample
and properly labelled, using supervised learning techniques. However, when we are confronted with new surveys in
which labelled data is scarce, we seek to adapt existing models from a resource-rich source domain to a resourcepoor target domain. Modelling the source domain to the target domain mapping can be done using different
methods and using several features which can be obtained from the time domain astronomy data.
Gene Expression Configurations of Signaling Pathway Components
Eeshit Vaishnav
Mentors: Michael Elowitz and Emily Capra
Cells respond to changes in their environment by sending, receiving and processing signals that originate outside
their borders. Signaling pathways often involve the use of multiple ligands, receptors and modulators. Recent
findings suggest that in some of these signaling pathways (like the Notch, which is a juxtacrine signaling pathway),
discrete qualitative behaviors emerge from the interactions of the multiple signaling pathway components. Singlecell RNA-seq is a powerful tool that potentially allows us to test these predictions about signaling pathway
expression patterns in-vivo. Here we conduct a systematic exploratory analysis of single-cell RNA-seq gene
expression datasets from varied sources ranging from immune cells to neural cells. We apply exemplar-based
unsupervised clustering methods (k-medoids) and dimensionality reduction techniques (t-SNE and PCA) on gene
expression data for signaling pathway components. These methods are implemented in the form of a robust
computational pipeline that is applied to single-cell RNA-seq data from two specific signaling pathways: the Notch
and the bone morphogenetic protein (BMP). We observe well-separated clusters in gene expression characterized
by archetypical expression patterns corresponding to each of these clusters. These archetypical expression patterns
are potential indicators of the different qualitative behaviors that the cells in these clusters display. Furthermore, a
complementary analysis in terms of evolutionary trade-offs, pareto-optimality and the geometry of phenotypic
space is then conducted to infer the biological tasks corresponding to these clusters in gene expression space.

77

Allosteric Inhibition of ERR by Pyrrole-Imidizole Polyamides in PTEN Deficient Cancer Cells

Daniel Valles
Mentors: Peter B. Dervan, Amanda Silberstein, and Tiezheng Jia
Pyrrole-Imidizole (Py/Im) polyamides are small compounds that can be programmed to attach bind to a variety of
DNA sequencesmotifs and, therefore, inhibit transcription. The Estrogen Related Receptor alpha (EER) is a nuclear
receptor in the P13K cell signaling pathway, which has been shown to lead to unregulated cell proliferation when
there is a lack of PTEN, a tumor suppressing gene. By targeting EER with polyamides, we hope to preventstop the
growth of cancer while limiting host toxicity. Four different PTEN-null cancer cell lines were used: AN3 CA, PC3,
786-O, and BT549 for human endometrial, prostate, kidney, and breast cancer, respectively. Two different hairpin
polyamides were used, both made to target EER but with different groups in the turn unit. The cells were dosed
with 10 different concentrations of each polyamide. The cytotoxicity of each polyamide on each cell line was
measured by determining the IC50 using the WST-1 assay after eithera 72 or 96 hour incubation period.
Invisible Catalyst Deposition on Micro-Wire Array Electrodes
Eugene Vaughan
Mentors: Harry Atwater and Sisir Yalamanchili
Photoelectrochemical cells capable of artificial photosynthesis are a potentially promising solution to the issue of
finding an alternative energy source in the coming years. This may be achieved by using sunlight to split water and
harvest the generated hydrogen gas as a fuel. We used silicon micro-wire arrays as an electrode for the cell, and
deposited catalyst on the tips of the wires in order to enable the hydrogen (or oxygen) evolution reaction. In order
to deposit catalyst only on the tips of the wires, we first deposited a layer of polydimethylsiloxane (PDMS) that
covered a portion of the length of the wires. When the catalyst was deposited, these covered regions of silicon
micro-wires were essentially guarded from the catalyst. The PDMS was then removed using an etching solution,
which also serves to remove the catalyst on the surface of the PDMS between the wires themselves. This means
that the catalyst is now deposited in a layer roughly 30 to 50 nanometers thick on the tips of the micro-wires, and,
due to the tapered shape of the wires, the metal catalyst does not substantially change the already-high absorption
rates of the wire arrays. This low level of alteration in the wire arrays reflectivity means that the catalyst is, in a
sense, invisible.
Half-Heusler Materials in Thermoelectrics
Victor Venturi
Mentors: Austin Minnich and Chengyun Hua
Thermoelectrics is a field of science that studies the direct conversion of heat and temperature differences to
electricity using solid-state materials. Thermoelectric devices create a voltage difference if two of their sides are
submitted to distinct temperatures. These devices do not have moving parts, which makes them useful for space
power generation and waste heat recovery. However, thermoelectric generators still suffer from having low
efficiencies. This makes their use restricted to cases where the only required feature from a generator is reliability,
rather that power generation. My project involves studying how half-Heusler materials perform in thermoelectric
devices compared to commercial materials. Half-Heusler materials have a huge potential for applications in
thermoelectrics. I redesigned, built and tested thermoelectric generators with both commercial materials and halfHeusler materials. I redesigned the generators to increase precision in measurements by changing the position of
the thermocouples. The building process was long and complicated, but I finally got around the issues with
soldering and was able to prepare the generator for testing, which was done subsequently. We hope that my work
will help scientists find a path to the solution for the low efficiency of thermoelectric generators.
Towards Molybdenum PNP Complexes With the Potential for Variable Binding Modes and Acid-Base
Non-Innocence
Ruomeng Wan
Mentors: Theodore Agapie and Kyle Horak
Fixation of N2 to ammonia is achieved industrially using the energy demanding Haber-Bosch process. In contrast,
molybdenumdinitrogen complexes have been shown to be able to catalyze N2 reduction under ambient conditions.
One of the most active Mo-based N2 reduction catalysts reported to date is a dimolybdenum-dinitrogen complex
bearing PNP type pincer ligand. From DFT calculation, it was proposed that the di-nuclear molydenum structure
plays a critical role during protonation in the catalytic mechanism. As a comparison, herein we report the synthesis
and characterization of mono-nuclear molybdenum complexes bearing a new PNP type pincer ligand to explore the
multiproton, multielectron reduction of N2. This PNP ligand has the potential to switch its binding mode to
molybdenum center from PNP pincer to P-pyridine system-P pincer, which may enable the pyridine nitrogen to
function as a proton relay during nitrogen reduction. We target Mo(N2) species by first metalating the PNP ligand to
form a Mo(CO)3 compound; and then via oxidation and photolysis, obtain CO-free molybdenum complexes, which
can potentially serve as precursors for Mo(N2) species following reduction.

78

Energy Harvesting: Using Aluminum-Carbon Redox Batteries as a Readily Available Source

Beiyi (Annie) Wang
Mentor: Hyuck Choo
The use of aluminum-carbon batteries as portable, cheap sources of energy is crucial in many endeavors, including
military explorations and third world country development. Through harnessing a simple redox reaction created
when aluminum and carbon are placed in a saline solution, this simple microsystem is capable of generating
enough power to most microdevices. We analyze how the materials perform under various conditions, optimize its
power output, and design a feasible microsystem. Size of the system is kept under 10 cm, and salinity varies from
9% to 35%, the salinity of body fluids to the salinity of seawater. Power generation varies by application and
consequently solution used, ranging from 7mA to 25mA. Success opens doors to the use of different materials for
implantation in to the human body or use in other regions lacking a steady supply of electricity.
Constructing a Y Chromosome-Linked X Shredder to Induce Population Crash Using CRISPR
Erin Wang
Mentors: Bruce A. Hay and Omar S. Akbari
Despite the development of vaccines, vector-borne diseases cause over one million deaths annually. Consequently,
vector control becomes a more desirable strategy for disease prevention. A genetic approach eliminates the threat
to the environment posed by insecticides and restricts the effects to the specific pest species. We are particularly
interested in developing a synthetic sex-ratio distortion system to induce population crash within a target species.
We currently have created lines of Drosophila melanogaster that express gRNAs targeting sequences unique to the
X chromosome. Once these are crossed with lines expressing Cas9 in the germline, we expect to observe Cas9mediated cleavage and shredding of the X chromosome during spermatogenesis, resulting in the production of
purely Y-bearing sperm. We then expect these flies to produce all-male progeny, leading to population crash. In
order to achieve drive beyond the first generation, we have produced constructs to be used to create Y-docking
sites on the Y-chromosome in Drosophila. The next step will be to use CRISPR to insert these docking sites, which
can then be used to link the X shredder to the Y chromosome.
Simulations of Systematic Effects in the nEDM Experiment
Jiahui Wang
Mentors: Bradley Filippone and Christopher Swank
The measurement of neutron electric dipole moments (nEDM), T and P odd interaction, is of great fundamental
interest. The possible existence of the nEDMs would be a signal of physics beyond the Standard Model and may
impact our understanding of the apparent matter-antimatter asymmetry of the universe. The study of systematic
effects plays a crucial role in understanding the sensitivity of the experiment. We use a Monte Carlo simulation
package to analyze the influence of geometric phase and different experimental conditions like temperature, the
influence of gravity, and the magnitude and shape of fields, which will provide an optimum parameter space for the
experiments and improve the overall precision.
Magnetic Manipulation of Cells for Brain Tumor Treatment
Michelle Wang
Mentors: Ali Hajimiri and Alex Pai
By varying localized magnetic fields, cells which have been loaded with superparamagnetic beads can be guided to
predetermined areas. Previous experiments using a 2D grid have shown success in aggregating and maneuvering
the beads themselves, and current experiments are being conducted to optimize the bead loading and movement
of various cell lines, including THP-1 and neural stem cells. This project focuses on the next step: 3D manipulation
of cells. Earlier work has shown that modified macrophages are capable of killing tumor cells; by injecting such
cells into the brain and guiding them to the tumor site with magnetic fields, we can treat tumors without the need
for invasive surgical techniques. To do this, we have begun development of a helmet that will be able to direct cells
from various angles. The helmet structure is made up of a set of 3D printed triangles which snap together to form
a dome. On top of every triangle is a coil through which current is run to generate a magnetic field. A flexible PCB
is affixed to each triangle to allow it to interface with adjacent triangles and be individually controlled. Development
of the system is ongoing, as are cell line experiments.

79

Implementing an Algorithm That Finds the Ground States of 1D Gapped Local Hamiltonians in
Polynomial Time
Shannon Wang
Mentor: Thomas Vidick
The density matrix renormalization group algorithm (DMRG) is used by condensed matter physicists over the world
to compute the ground states and the ground state energies of different Hamiltonians and models. However, the
DMRG may get stuck in non global minima; there is also no guarantee that it will converge to a ground state within
polynomial time. A recently published paper in Nature Physics by Landau et al. offers a proof of a new algorithm
that finds the ground states of 1D gapped local Hamiltonians in polynomial time. Here, the results of the
implementation of the algorithm for frustration free and frustrated Hamiltonians are discussed and compared with
the results yielded by the DMRG for the same models.
DNA-Based Competitive Fuel Implementation of the Winner-Take-All Function
Siyuan Stella Wang
Mentors: Lulu Qian, Kevin Cherry, and Robert Johnson
Since its discovery, DNA has been proven to be not only the source of genetic information but also an effective
building material in programmable nanodevices and soluble circuits for applications in targeted medical delivery
and control of cell systems. DNA motifs for the linear threshold model of the neuron can be used to compose
tunable neural networks capable of pattern-recognition and completion. These circuits can be simplified using the
winner-take-all (WTA) function. My project seeks to develop and experimentally test a WTA implementation with
desirable qualities such as constant number of toeholds, linear increase in system size, and complexes of no more
than two strands. Competition imposed by a limited shared fuel helps to achieve the linear size, and use of a
previously developed seesaw DNA motif allows for a constant number of toeholds with two-stranded complexes
while using a reliable motif. The project consists of understanding published theoretical, developing a new system
with desirable properties using that understanding, simulating its behavior, and testing its behavior experimentally.
Through this investigation, we can learn how effective fuel-driven competition is at minimizing circuit size and
possibly develop a WTA motif unit for more complex computation.
Detection of Radio Transients Using the Owens Valley Long Wavelength Array (LWA)
Yuankun Wang
Mentors: Gregg Hallinan and Marin Anderson
The LWA at Owens Valley is a radio interferometer capable of imaging the entire sky every 9 seconds at 10-88MHz,
making accessible previously unexplored timescales and frequencies. Objects emitting in this bandflare stars,
brown dwarfs, and possibly hot jupitersare bright due to synchrotron or coherent radiation rather from than
thermal radiation, and are thus quite faint or unobservable at visible wavelengths. I have developed a pipeline for
the LWA to detect these, and other, radio transients. Using data from a 10 day run of the LWA, I find that
variations in flux as sources pass through the primary beam of the array, and from sidelobes of the brightest
sources can be fitted out by dividing data at a period of one sidereal day. The remaining rms noise is on average
7.19 Jy for resolvable sources between 50-1000 Jy: theoretically sensitive enough to detect a ~10 Jy flare. For
unresolved sources, the noise averages out to ~280 Jy when summing fluxes over a 1.25 degree aperture. Some
observed variations in this normalized data are thought to be due to ionospheric interference. Addition of long
baselines to the LWA will increase resolution from ~1 to 7 and imaging in circularly polarized radiation will reduce
noise due to confusion while increasing the number of resolvable sources, increasing the likelihood of detecting a
transient.
2-Selmer Ranks of Quadratic Twists of (Hyper)elliptic Curves
Matthew Weidner
Mentor: Majid Hadian-Jazi
The study of ranks of elliptic curves is a classical and difficult Diophantine problem. The 2-Selmer rank provides an
upper bound on rank which is effectively computable and better understood. There has been much recent work by
mathematicians like Klagsbrun, Mazur, Rubin, etc. trying to understand how the 2-Selmer rank varies in families of
elliptic curves called quadratic twists. In this project, we apply some of these new results to show that the parity of
the 2-Selmer rank is periodic in families of quadratic twists, with a period that is easily computed. Assuming a
standard conjecture, this allows one to construct infinite families of positive rank quadratic twists of most elliptic
curves, including all elliptic curves over number fields with a real embedding. In particular, we obtain a new proof
for a known result concerning the famous congruent number problem. We also generalize these results to
Jacobians of odd-degree hyperelliptic curves.

80

Mechanistic Investigations of Cobalt Glyoximes

Lucille Wells
Mentors: Harry Gray and Sarah Del Ciello
Using the suns energy to split water and store fuels has been proposed as an alternative to the fossil fuels used
today. Earth-abundant catalysts are needed to facilitate the reduction of protons to form hydrogen gas in the
water-splitting process. Cobalt glyoximes are known to be catalysts for hydrogen evolution; however, the
mechanisms require further investigation. This study aimed to address whether the mechanism of the cobalt
glyoxime was homolytic or heterolytic through the principle of microscopic reversibility. In addition, the
investigation of cobalt glyoximes led to a focus on finding a protonated intermediate in hydrogen evolution by
reducing the cobalt glyoxime and protonating it with a weak acid.
The Characterization of Ncad and Ecad Expression in Murine and Melanoma Cancer Cells
Samuel Wie
Mentors: Marianne Bronner and Crystal Rogers
Cadherins are transmembrane proteins that are critical for forming adherens junctions, which bind cells within
tissues together. Changes in cadherin expressions are associated with with the epithelial to mesenchymal transition
(EMT) during embryonic development and cancer metastasis. The goal of this project is toHere we used
immunohistochemistry and Western blot analysis to identify, characterize, and analyze the expression of Ncadherin (Ncad) and E-cadherin (Ecad) cadherin proteins during EMT in developing mouse embryos, and to
compared these data to those the expression of both cadherins from the expression in metastatic cancer breast
cancer (MCF7) and mouse and human melanoma cells (B16, M202 and A375). Immunohistochemistry, western
blots, andWe find that in murine trunk neural crest cells, similar to migratory metastatic melanoma cells, Ncad is
expressed on the cell membrane while Ecad is absent, which is the opposite of the expression profile in epithelial
breast cancer cells. We performed gain of function studies transfections were utilized in melanoma and breast
cancer cells to analyze the effects on cadherinlocalization of the dominant cadherin s from overexpressing a specific
cadherin via transfection or knocking down a specific cadherin via the use of CRISPR/cas9after mis-expression of
the lacking cadherin proteins. Proximity ligation assay will be used to determine if these cadherins can interact
heterophilically in melanoma and murine cells. A western blot shows that mesenchymal B16 has a lot of Ncad but
no Ecad, and epithelial MCF7 has a lot of Ecad, but no Ncad. Following transfection with full l and determined that
ength Ncad, MCF7 human breast cancer cells showed a considerable decrease in Ecad and a significant increase in
Ncad. We conclude that adding exogenous Ncad or Ecad reduces the membrane associated levels Ecad and Ncad,
respectively, due to the lack of available intracellular cadherin-stabilizing components such as -catenin.
Smart Needles for Delivering Chemotherapy in a Medical Patch
Theodore Wilkening
Mentor: Mory Gharib
The objective of this project is to study the use of gelatin, albumen and other biocompatible compressible
substances as tools for guiding flow through needles to be used in chemotherapy patches. In this experiment, the
resistance to the exit flow of the different needles was controlled, similar to how skin offers different resistances
due to the layer and thickness of the layer of skin. The goal is to guide fluid through needles with less resistance.
These less resistant areas represent the subdermal layers of skin where the chemo can be distributed throughout
the body. Results have shown that with equal amounts of gelatin and with the absence of air between the gelatin
and fluid, the fluid flow is guided towards the less resistant path while the more resistant path remains clogged. A
viable option for mechanically guiding flow to the proper location in a chemotherapy patch, further research must
be done on consistently filling needles with equal amounts of compressible material.
Design and Fabrication of CMOS Electrochemical Sensors for In Vivo Detection of Cancer Metabolites
and Cellular Signaling Molecules
Anna Winnicki
Mentors: Axel Scherer, Muhammad Mujeeb-U-Rahman, and Deepan Kishore Kumar
To date, electrochemical detection is the only available technique sensitive enough to detect physiologically
relevant concentrations of signal transduction molecules in real time and in vivo. Modification of the electrode
surface affords measurable current signals for analysis using various electrochemical techniques, such as cyclic
voltammetry and amperometry. Electrochemical methods have been used to detect nitric oxide (NO), a cellular
signaling molecule of cancer, and lactate, a metabolite found in most tumors. Herein, we seek to optimize
electrochemical detection of NO and lactate by developing electrodes with improved sensitivity towards target
molecules. Using COMSOL Multiphyiscs, a finite element method solver, in conjunction with MATLAB, an optimal
photomask design was produced for photolithography. Electrodes were fabricated according to the photomask

81

design and testing was performed in solutions of lactate and nitric oxide produced by decomposing S-nitrosoacetyl-DL-penicillamine (SNAP). Moving forward, we plan to continue refining the electrode design and fabrication
process. Developing electrodes that detect cellular signaling molecules impacts our understanding of cancer
biochemistry and cancer prognosis and could potentially aid in monitoring and evaluating the therapeutic effects of
chemotherapy.
Characteristics of Observed Seismic Noise Due to Sediment Flux
Bismark Wong
Mentors: Victor Tsai and Florent Gimbert
Measuring sediment flux in rivers is typically time-consuming and complicated. Previous models attempt to the
predict flux of the particles impacting a riverbed from the recorded Power Spectral Density (PSD) of a nearby
seismic station. By utilizing the PSD with river parameters such as depth, sediment diameter, and distance to the
station, an estimate of the sediment flux can be obtained. In this paper, station responses that exhibit modeled
characteristics are documented and analyzed to determine model relevance. We also discuss unpredicted effects
and observed phenomena including correlation of PSD to river discharge, hysteresis, and possible fluvial response
caused by water discharge. If the model is statistically significant for sediment or fluvial response, such methods of
estimation may be used in place of sediment traps to gather flux data. In addition, it serves as basis for landscape
evolution models where evolving river channels, erosion and sediment transport are factors.
Zinc Nanowire Substrate Fabrication for Glucose Detection Using SurfaceEnhanced
Raman Scattering (SERS)
Kelly Woo
Mentors: Hyuck Choo and Daejong Yang
Current technology requires diabetics to undergo painful, inconvenient, and discontinuous measurement processes
several times a day. Thus, we hope to develop more convenient ways of measuring glucose by utilizing surface
enhanced Raman spectroscopy (SERS) technique. To accomplish commercially viable SERS technologies for glucose
detection an optimal substrate must be designed with higher electromagnetic enhancement. Many variables must
be taken into consideration, such as nanoparticle sizes, shapes, dielectric environment, and resonance of the
designed structures. To create these substrates, we chose to hydrothermally grow zinc nanowires on silicon wafers,
and then deposit gold nanoparticles. My objective was to manipulate the synthesis process to produce controlled
zinc nanowire growth on a substrate to significantly enhance the signals produced by Raman spectroscopy. To
change the morphology of the zinc nanowires, parameters, such as the pH of the solution in which the nanowires
are grown, the temperature during growth, and the ratio of compounds composing the solution itself, were
changed. It was found that in more concentrated solutions, zinc nanowires grew with greater diameter. I was also
able to consistently grow pencil/cone shaped nanowires. Upcoming research includes fabricating a very uniform
substrate and determining which nanowire structure will yield the greatest SERS signal.
Genome Editing to Address Problems in Biomanufacturing
Yuanyuan (Tina) Xu
Mentors: Kelvin Lee, Jongyoun Baik, and Justin Bois
Chinese Hamster Ovary (CHO) cells are a major mammalian host cell for biomanufacturing of therapeutic proteins,
and they have variable expression of host cell proteins (HCPs) that could prove detrimental to humans if present
along with therapeutic proteins. Several HCPs have been found to be difficult-to-remove using common protein
purification techniques. We hypothesized that with a sequenced genome and identified difficult-to-remove HCPs, it
would be possible to use the CRISPR/Cas9 technique to minimize or eliminate the expression of HCPs. From a list
of ten previously identified difficult-to-remove HCPs, we chose to target Cathepsin D, Nidogen-1, and Prosaposin.
We then transfected CHO cells with a Cas9 plasmid and a plasmid containing a single guide RNA (sgRNA) sequence
found through the online sgRNA tool CRISPy. Then, the cells will be tested for genome expression using a T7
endonuclease assay and enriched. We would then test the cells for HCP expression with western blots. This study
will assess the feasibility of using CRISPR/Cas9 on CHO cells to modify their genome sequences and deal with HCPs
in the future.
Arithmetic Topology: Bost-Connes Quantum Statistical Mechanical System of Branched Covering of
3-sphere S3
Yujie Xu
Mentor: Matilde Marcolli
This project arises from a question posed by Masanori Morishita, in relation to his work on arithmetic topology. The
basic idea of arithmetic topology is a dictionary of analogies between number fields and 3-manifolds, that make it
possible to find analogs of number theoretic results in topology and vice versa. The Bost-Connes system and its
generalizations associate to a number field a quantum statistical mechanical system, whose partition function is the
Dedekind zeta function, whose symmetry group is the abelianized absolute Galois group, and whose ground states
are related to abelian extensions of the field. Morishita asked whether it is possible to similarly associate to a

82

compact oriented 3-manifold a quantum statistical mechanical system which encodes the information on certain
abelian branched coverings, and what would be the meaning of the corresponding time evolution and partition
function. The aim of this project is to investigate this question by explicitly constructing such systems using
branched covering data and study their properties.
Jupiter Hildas Two Distinct Color Popluations and Magnitude Distribution
Yixiao Yan
Mentor: Michael E. Brown
The Hildas are a group of significant minor bodies near Jupiter in the middle Solar System that have mysterious
origins. Their orbits form a triangular region near Jupiterber theoretic results in tasteroid belt. Several recent
studies have shown that these objects revealed bimodalities with respect to near-infrared spectra, infrared albedo,
and color. Asteroid information from the Minor Planet Center catalog and color data from the fourth release of
Sloan Digital Sky Survey indicate that most Hilda asteroids can be categorized into two distinct groups in terms of
spectral slopes: the less red populations, and the red populations. In this paper, we analyze the color populations
by creating a histogram and doing Gaussian fit lines, and compare and contrast these color data to Jupiter Trojans,
which have proven to have different peaks of less red and red populations from Hildas. We will also demonstrate
the magnitude distributions of two colors populations of Hildas, and fit them to a broken power law. It is possible
that the less red and red populations of Hildas come from different origins, and may or may not resemble those
two categories of Jupiter Trojans. We propose a few hypotheses for the origin and evolution of the Hilda population
based on analyzed data, but it remains unknown whether Nice Model is correct or not.
Physics-Based and Empirical Models of Site Response for SCEC Ground Motioning Simulations
Albert Yang
Mentor: Domniki Asimaki
Existing models for site-specific ground motion analyses intended for earthquake simulations fall into several
categories based on the method of solving. The linear and equivalent linear methods provide a quick approximation
of site response, in contrast to more rigorous solutions such as incremental nonlinear methods. For this SURF
project, three modules have been developed for the Southern California Earthquake Center (SCEC). The modules
will be shared with the seismological and engineering members of SCEC through SCECs computational platform for
earthquake simulations. These programs will take in an input motion and generate a surface motion for low to
medium input accelerations at soft soil sites prone to large ground deformations.
More specifically, the first two programs utilize a well-known, classic wave propagation model that takes an input
motion and utilizes transfer functions to return a surface motion. The last program introduces a modification to the
classical equivalent linear algorithm to correct for over-dampening and deformations at higher frequencies. By
creating an upper cut-off frequency for the strain frequency spectrum, and modifying the least squares fit
accordingly, high frequencies are not filtered out as strongly as in the original method, and manifest themselves
accordingly in the calculated site response.
Future work to be done may include the further optimization of the equivalent linear algorithms in Python, in terms
of both accuracy and speed.
Quantification of Stable and Unstable Wavefront Propagation in Ultrathin Films Subject to
Marangoni Forces
Cassidy Yang
Mentor: Sandra M. Troian
Molecular monolayers, used to reduce the surface tension of the liquid on which they are deposited, will typically
spread over the entire liquid surface until the interface is characterized by a constant surface concentration.
Marangoni forces responsible for this phenomenon generate complex traveling waves in an effort to minimize
gradients in surface concentration on the way toward equilibrium. In this study, we have built a monochromatic
interferometric system to quantify wave front propagation in ultrathin liquid films subject to Marangoni forces. This
imaging technique is particularly suited to this study since these traveling waves exhibit significant variation in
surface slope, which is highlighted by the distortion of interferometric fringes. From analysis of the distortion
patterns, we aim to deduce the spatiotemporal behavior of several moving fronts, indicative of stable and unstable
propagation. The speed of these small amplitude waves as a function of time will then be directly compared to
theoretical predictions based on the hypothesis that the traveling waves exhibit self-similar dynamics during
propagation.

83

Predicted Structures and Odorant Binding for the Human Olfactory Receptors OR51E1-1, OR51E2-1,
and OR51L1-1
Sherrie Yang
Mentors: William Goddard, III, and Soo-Kyung Kim
Olfactory receptors (OR) are protein receptors in the nose that help determine the sense of smell by binding with
odorants. In humans, ORs are separated into two classes: Class I (fishlike receptors) and Class II (tetrapod-like
receptors). Not much is known about the structure and binding preference of ORs, but we can computationally
predict the best structure of ORs by analyzing samples of millions of conformations and selecting the best structure
through a method called GPCR Ensemble of Structures in Membrane BiLayer Environment (GEnSeMBLE). Using
hOR1G1 as a template, the most likely structures of the fish-like genes OR51E1, OR51E2, and OR51L1 can be
found. Binding preferences can be studied for each OR using known odorants and preferred ligands through the
DarwinDock method. Accurately identifying ligand binding sites in proteins can help optimize the ligands with
desirable properties.
Using Space-Based Geodesy to Monitor Changes in Shallow Aquifers
Shuo Yang
Mentor: Mark Simons
Rapidly increasing social demand of water resources has caused growing concern of monitoring changes in the
shallow aquifer systems, especially in the southern California of the United State. We apply GIAnT software suite
and develop tools in Python to facilitate analysis of the behavior of evolving shallow aquifer systems using data of
the Interferometric Synthetic Aperture Radar (InSAR). Raw data is acquired by the European ERS 1/2 and Envisat
satellites from June 1992 to September 2010, with a particular emphasis on the Los Angeles Basin and the
Newport-Inglewood Fault Zone. Visualization of the result gives out both amplitude and timing pictures of the
periodic components of the InSAR time series analysis using sinusoids, demonstrating a close relationship with
aquifer geometry and spatial variation, which allows for a better view of the effects of boundaries structures of the
aquifers on the local hydrologic systems.
User Interface for Developing Streaming Apps: Stream Programming for Novices
Ker Lee Yap
Mentor: K. Mani Chandy
With only basic Python functions and a rough understanding of the open source PStreams package, beginning
programmers can easily and quickly build streaming applications using this tool. Users simply choose from a
collection of simple pre-made functions from common libraries and specify the type of PStreams wrapper they
want. Features include a drag-and-drop interface where modules can be connected and an animated debugger that
displays data as it passes through the network. Users can also bypass the graphical interface entirely and use this
tool to execute JSON files describing the graph. Besides single process streaming applications, this tool is suited for
developing apps with multiprocessing and cloud computing.
The Effect of Ablation of Lin28 and let-7 on R-848-Mediated B Cell Responses
Chittampalli (Yasha) Yashaswini
Mentors: David Baltimore and Shuai Jiang
let-7 is a highly conserved microRNA with a variety of functions, but is most well-known for its tumor suppressor
function. Lin28 is a protein that is involved in development, tumorigenesis and metabolism, and has a doublenegative feedback loop with let-7. The objective of this project is to investigate the reaction of splenic B cells from
let-7 knockout mice and Lin28 knockout mice to R-848, a drug with potent antiviral and antitumor activities.
Through R-848 treatment of splenic B cells from let-7 knockout mice and Lin28 knockout mice, and analysis with
FACS and ELISA, we have observed that absence of Lin28a and Lin28b results in decreased R-848-mediated B cell
activation, and that absence of let-7b and let-7d results in increased B cell activation. We have also observed that
Lin28b knockouts have decreased antibody secretion, and let-7b and let-7d knockouts have increased antibody
secretion. These results suggest that Lin28 and let-7 play roles in R-848-mediated B cell response, and that let-7b,
let-7d, and Lin28b are involved in antibody diversification via class switch recombination. Suggestions for the
continuation of this project include conducting qPCR and Western Blots to identify which class switching genes and
proteins are regulated by let-7 and Lin28.

84

Exciton Bound States in Topological Insulators

Kexin Yi
Mentors: Liang Fu, Vladyslav Kozii, and Jason Alicea
In this report, we study single exciton bound states on top of an isotropic conduction and valence and structure
with opposite parity, which commonly exists in topological insulators. By integrating out the angular part of the
two-body wave function, the Schrodinger equation is turned into an eigenvalue problem on a one dimensional
radial momentum space. Exciton binding energies can be calculated numerically after proper discretization of the
model. We show that in the band inversion regime, the binding energy of the lowest exciton branch is significantly
enhanced compared with the uninverted regime, leading to easier formation of an exciton condensate. In the limit
of weak spin-exchange, our results suggest a hidden chiral order in the exciton condensation phase, breaking
parity while preserving full rotation symmetry.
Upper Bounds for Non-Crossing Planar Graphs
Chenchao You
Mentor: Adam Sheffer
For a set P of n points in the plane, we are interested in an upper bound for the number of various crossing-free
graphs that can embedded over P, including general planar graphs, spanning cycles, and matchings. Previous
methods mainly relied on the triangulations of P. To improve upon those previous results, we perform a vertical
decomposing a planar graph and define the rank of vertices and edges in it. By only removing and inserting edges
of low rank, we obtain a recursion for the number of graphs with respect to the number of the edges in them.
iPS Cell-Derived Macrophages Function in Fibrodysplasia Ossificans Progressiva
Kristie Yu
Mentors: Edward Hsiao, Emilie Barruet, Marcela Morales, and Elaine Hsiao
Heterotopic ossification, the abnormal formation of endochondreal bone in soft tissues, is often triggered by trauma
and can cause significant disability and pain. Although the immune system normally plays critical roles in mediating
responses to tissue trauma and injury, and can significantly impact the skeleton in many clinical settings, we
understand little about the mechanisms that link innate immune cells and skeletal function, particularly in humans.
Human induced pluripotent stem (iPS) cells provide an opportunity to create human-specific models of skeletal
diseases. Our work focuses on fibrodysplasia ossificans progressiva (FOP), a rare genetic disease. Patients with FOP
are extremely sensitive to injury or inflammation, which can induce rapid and dramatic bone formation. In this
study, we hypothesized that abnormalities in the innate immune response leads to such heterotopic ossification. To
elucidate this mechanism, we have optimized a protocol to create iPS cell-derived human macrophages, a type of
white blood cell that is a key regulator of the innate immune response. iPS cell-derived macrophages from both
FOP and wild-type patients were characterized by FACS. We found that we can create CD11b+ cells (macrophages)
with up to 80% yield, though there was high variability among individual differentiation experiments. Among the
CD11b+ cells, there were CD163+ (~60%) and CD163- (~40%) cells, which suggests that we obtain both M1- and
M2-type macrophages that function in inflammation and tissue repair, respectively. These cells will help us
understand the role of macrophages in FOP and other diseases of heterotopic ossification, and serve as a
foundation for studying other macrophage conditions.
High Quality Growth of MoS2 for Improved Sensitivity and Selectivity in Chemical Sensing
Patrick Yu
Mentors: Brent Keller, Jeffrey Grossman, and Harry A. Atwater
Chemical sensing is a technique that can provide information about the chemical composition of the environment in
gas or liquid phase and has been used in wide variety of fields such as medicine, agriculture, industry, transport,
environmental control, and other fields. Recently, transition metal dichalcogenides (TMDs) have caught the
attention of researchers attempting to improve chemical sensing capabilities due to their high surface area to
volume ratios and atomic thicknesses. Monolayer MoS2, which is a TMD, has been recently observed to retain high
sensitivity to gas phase exposure of a few analytes, while also featuring enhanced selectivity1. The high sensitivity
of the monolayer MoS2 is due to its intrinsic small dimension which can provide single-molecule level information.
High quality MoS2 samples are prepared by etching MoO3 using argon gas or hydrofluoric acid, using chemical
vapor deposition to sulfurize the MoO3 into MoS2, and transferring the MoS2 onto a new Si/SiO2 wafer. Atomic force
microscopy (AFM) is used to determine the thickness of the MoS2. Photoluminescence (PL) and Raman
spectroscopy are used to determine the electronic bandgaps of the MoS2 samples. Since it has been observed that
MoS2 transforms from an indirect to a direct bandgap semiconductor as it becomes more monolayer, tuning the
thickness of MoS2 tunes its electronic bandgap. By tuning the electronic bandgap to determine high quality MoS2
samples, better quality chemical sensors can be fabricated. This will result in improved sensitivity and selectivity
for an environments chemical composition.
REFERENCE:
Perkins F.K. et al. Chemical vapor sensing with monolayer MoS2, Nano Lett. 13 (2013), 668.

85

Creating a Consistent History of Galaxy Evolution

Dominic Yurk
Mentors: Peter Capak and Charles Steinhardt
One of the great challenges of observational astronomy is that we can only see snapshots of galaxies; there is no
direct way to see their evolution. In the past it has generally been assumed that galaxy evolution is a chaotic
process with many different paths, making it effectively impossible to determine the history of a galaxy based on
observations of its current state. However, recent observations have revealed that almost all active galaxies lie on
a main sequence, giving us a tight relation between galactic age, mass, and star formation rate. The correlation
between these parameters is very tight with few outliers, implying that most galaxies do in fact follow similar
evolutionary paths. In this paper we combine observations of the main sequence with simulations of galactic
merger rates to create a relatively simply evolutionary track for galaxies based only on their initial mass and
formation time. To test the validity of our model we evolve a population of galaxies along this track and find that
the results are consistent with observed galactic mass distributions. This galactic evolution track will allow
astronomers to extrapolate backwards from the current state of a galaxy to determine its history, opening up new
possibilities for analyzing how galactic populations interact and change over time.
Design, Assembly, and Testing of a Higher Performance Gas Pressure Launching System for Lindurst
Laboratorys 40 mm Propellant Gun
Brian Zdeb
Mentors: Paul Asimow and Michael Burns
The 40 mm projectile launcher in Lindhurst Laboratory can currently only use gas pressures up to 300 psi for some
shots, and must use gun powder to achieve higher velocities. It is our task to design a new system to deliver high
pressure gas to a 0.25 kg projectile to reach a maximum velocity of 1500 m/s, so that gun powder will not be
necessary for future experiments. Achieving these upper limits would require containing at most 10000 psi in a
roughly 5L container before releasing it instantaneously to the projectile. We propose a design that is essentially a
custom built dump valve incorporated into the already existing breech vessel. A front and back piston, differing in
areas, provides an opening force to unplug the smaller front piston from an orifice leading to the projectile.
Electromagnets will be used to hold the dump valve closed. When turned off, they will release the rear piston,
which pulls the front piston with it, resulting in a shot. Material acquisition, fabrication, and testing of the design
are the next steps for this project to be completed.
Context-Dependent Flight Responses to Small Objects in Tethered Drosophila melanogaster
Sasha Zemsky
Mentors: Michael Dickinson and Floris van Breugel
During flight, Drosophila melanogaster may change their responses to particular visual stimuli under different
conditions to optimize the chances of finding food, finding mates, and avoiding predators. In free flight, Drosophila
normally turn away from small, blob-like objects, but in the presence of an attractive odor, they are more likely to
approach these objects. Flies continue to exhibit blob attraction several seconds after leaving an odor plume,
suggesting that olfactory and visual memory may play important roles in food search. It is difficult to track the
precise time course of the flies response to odor in free flight, so we developed a tethered flight arena to study this
behavior. We presented flies with different combinations of visual stimuli and recorded their turning responses both
with and without odor. The results unexpectedly show that flies responsiveness to blobs declines as the number of
blob presentations increases, suggesting that they can use recent visual memories of objects to alter their
behavior. This phenomenon provides a simple paradigm to study insect visual memory in future experiments. In
addition, we can use this information to design experiments that maximize flies responsiveness to blobs over time
to produce a more robust odor response.
Robotic Laser-Adaptive-Optics Imaging of Binary and Multiple Star Systems
Celia Zhang
Mentors: Reed Riddle and Lynn Hillenbrand
Around half of all stars are believed to be in binary or multiple star systems. The insight multiplicity gives us on
star information is crucial in understanding the evolution of matter in the universe. Robo-AO is an instrument
mounted on the Palomar 1.5m telescope that uses adaptive optics to quickly and autonomously image astronomical
objects at the native spatial resolution of the telescope. Images gathered by the Robo-AO instrument were
examined to identify and characterize binary systems in the Pleiades, Praesepe, and NGC 2264 clusters. Using
photometry software, the relative positives and magnitudes between each pair were determined. Software was
developed to correct for instrument distortion. The point-spread function of each primary star was modeled using
both a Gaussian and Moffat function. Contrast curves were generated to determine the completeness of the
binarity detections. Further analyses are to be done to conclude multiplicity and binary star properties.

86

Neural Circuitry Underlying Sun Compass Navigation in Drosophila

Tiffany Zhang
Mentors: Michael Dickinson and Peter Weir
When skylight is scattered by particles in the atmosphere, its electric field oscillations become linearly polarized in
an orientation dictated by suns position. Behavioral experiments have shown that animals can detect the resulting
pattern and use it to estimate the position of the sun; specifically, Drosophila melanogaster use this ability to orient
flight when other cues are unavailable. Our goal is to identify the neural circuitry underlying this behavior, starting
with candidate neuron classes identified by previous brain imaging with genetically-encoded calcium indicators. We
monitored flight by placing a fly in a custom-built arena that introduced rotating polarized UV light and
simultaneously recorded the flys wing-stroke amplitudes, which we used to infer steering maneuvers. After
verifying that wild type fruit flies respond to polarization, we used the GAL4/UAS system to genetically silence each
neuronal class and monitored the effect on behavior. We first silenced photoreceptors containing the UV-detecting
opsin rh3 and observed a loss of polarization vision, and preliminary data suggests that flies with certain neuron
classes in the optic lobe silenced are also affected. With this work, we hope to better understand the
implementation of neural computations in fruit flies and in other organisms.
Platform-Based Design for Robotics Applications
Yuening Zhang
Mentors: Richard Murray, Vasumathi Raman, and Samira Farahani
Platform-based design is a systematic design methodology that decomposes a complex problem into different
layers and achieves modularity with the components. It can potentially reduce the development cycle and promote
reuse and optimization of functionalities. However, there has not been a fully established platform-based design
framework for many robotics problems, both because computational tools for robotics problems still have
limitations and the robotics design today are mostly specialized for the problem. This project will explore platformbased approaches for motion planning problems using the Temporal Logic Planning Toolbox (TuLiP), BluSTL and
Robotic Operating System (ROS). The project started with defining an example problem, and constructing a
platform architecture that segregates the problem formulation. The problem is then implemented by utilizing the
above toolboxes, which includes developing functionalities to connect the toolboxes. If simulation works correctly,
the controller will be taken to a kobuki-based turtlebot in NCS lab. In the end, a generalized platform architecture
will be proposed for motion planning problems.
Study of Quasiparticle Lifetime in MKID-Based Dark Matter Detector
Songyuan Zhao
Mentors: Sunil Golwala and Brett Cornell
Resolution of MKID-based dark matter detector has a -1/2 quasiparticle lifetime (tau) dependence. By studying
behaviour of quasiparticle lifetime one hopes to increase the overall sensitivity. It has been shown by previous
studies that for MKID in a light tight box, tau plateaus at low temperatures. This project studies the behaviour of
tau against temperature for MKID devices previously fabricated by Dr. David Moore. Measurements of tau are done
by fitting generation-recombination (g-r) noise pattern with theoretical model. By considering established theories
of g-r noise, it has been shown that phase direction of g-r noise should be visible above the HEMT amplified white
noise. Analysis of existing data has shown that TLS noise could potentially overshadow g-r noise in our devices. In
this paper, we present both our theoretical analysis of various noise factors, as well as data from devices which
display the highest amount of g-r noise, before concluding with the temperature dependence of quasiparticle
lifetime.
Optimizing Platinum Nanoparticle Size for Use in Solid Acid Fuel Cells
Christopher Zhen
Mentors: Sossina Haile, Sara Thoi, and Hadi Tavassol
Solid acid fuel cells (SAFCs) are a new class of fuel cell with promise in the clean conversion of chemical fuels to
energy. Because of their higher operating temperatures compared to polymer electrolyte membrane fuel cells
(PEMFCs), SAFCs are potentially more efficient and less sensitive to fuel impurities. This means that methanol, a
much cheaper alternative to hydrogen gas, can be used as fuel. The goal of this project is to improve the efficiency
of SAFCs by reducing the size of platinum nanoparticles used in the catalyst. Reducing particle size increases the
surface area of the platinum catalyst and allows the existence of more triple-phase boundariesareas where the
catalyst, gas fuel, and electrolyte are all in contactincreasing catalyst efficiency. In this research we first found a
way to successfully control the size of the platinum nanoparticle by modifying the synthesis temperature and then
established a relationship between catalyst activity and Pt nanoparticle size.

87

Towards Single Phonon Fock State Generation in an Optomechanical Crystal

Hengyun (Harry) Zhou
Mentors: Oskar Painter and Gregory MacCabe
Quantum information processing requires the generation and detection of non-classical quantum states and the
ability to interface disparate quantum systems to utilize their respective advantages. Optomechanics provides a
natural interfacing platform in which the generation of Fock states would be important progress. We analyze the
experimental requirements for successful generation, heralding, and subsequent detection of a single phonon Fock
state in an optomechanical crystal and describe ways to improve our system for optimal signal-to-noise ratio in a
Hanbury-Brown Twiss measurement of phonon statistics. In particular, we highlight ways to engineer the
mechanical thermalization rate by varying the period number of a phononic shield surrounding our device, so that
pulsed experiments can be performed at higher repetition rates, significantly reducing the acquisition time required
for single phonon Fock state generation. These results are confirmed both numerically and experimentally, and
could pave the way to future quantum information processing tasks in optomechanical systems.
RAG-RSS Binding Interactions Through Tethered Particle Motion
Haojie Zhuang
Mentors: Rob Phillips and Soichi Hirokawa
V(D)J recombination is crucial for the diversity of antibodies produced by our immune system. The first step of this
process involves binding of recombination activating gene (RAG) protein to recombination signal sequences (RSS),
then under the right conditions, the formation of the paired complex loop. It has been shown that this process can
be observed at the single molecule level through tethered particle motion (TPM) experiments, in which a bead is
held by a strand of DNA to a slide and allowed to undergo restricted Brownian motion. The length of DNA free to
move correlates with the beads motion in the horizontal plane, allowing the measurement of the effective length of
DNA by this motion. This effective length is changed when a protein binds to the DNA, and it is this observation
that allows the measurement of dissociation constants and dwell times of this reaction. Further studies are aimed
at quantifying the parameters of binding energies during this process to draw up a statistical mechanical model of
the RAG-RSS interaction involving full-length RAG proteins and nonconsensus RSS, as well as creating a standard
procedure for tethered particle motion experiments.
Kernelized Dueling Bandits
Vincent Zhuang
Mentors: Yisong Yue and Yanan Sui
We study the Dueling Bandits problem, a variant of the multi-armed bandits problem with preference feedback,
under the additional assumption that there exist inter-arm relationships encoded in some kernel. More specifically,
we assume that either a latent utility function or the preference matrix is sampled from a Gaussian process or
alternatively a corresponding function in the reproducing kernel Hilbert space. We propose several novel algorithms
for this problem, including GP-Doubler and Continuous Doubler, and provide theoretical guarantees in the form of
expected and PAC regret bounds. We also demonstrate empirical performance on both synthetic and real data.
Investigating Supersymmetry Through Combinatorics
Nick Zolman
Mentor: Matilde Marcolli
An Adinkra is a discrete finite graph that encodes all the information about an off-shell supersymmetry algebra. We
investigate how supersymmetry may be linked to other graphs by looking at the relationship between Adinkras and
dessins denfants (a type of bipartite graph embedded on a Riemann surface). We also look at various graph
products involving Adinkras in an attempt to embed supersymmetry into a larger algebra. Our investigation leads
us toward the study of origamis and origami curves (a finite covering map of a torus) as a hopeful method of
investigating supersymmetry further.

88

Solid-State Tunable Optical Metamaterial Surface

Leo Zornberg
Mentors: Harry Atwater and Krishnan Thyagarajan
The principle objective of this project was to fabricate a solid-state tunable optical metamaterial surface; a surface
whose absorption spectra can be reversibly altered post-fabrication. We propose to accomplish this by using the
phenomena of gating, whereby a transparent conducting oxide (TCO) exhibits a charge carrier accumulation layer
when an external voltage is applied, in order to alter the plasmonic resonances of a patterned metallic surface. In
this project, we fabricated an inverse dolmen structure in silver using focused ion beam (FIB) lithography, growing
a thin insulating film of aluminum oxide using atomic layer deposition (ALD), and lastly depositing a layer of indium
tin oxide (ITO). The reflective properties are to be investigated using VIS-IR spectroscopy. In simulation, an
externally applied voltage was demonstrated to suppress and blue-shift the location of the absorption peaks, and
we hope to experimentally verify these simulations in the upcoming weeks. As such, we have demonstrated a novel
design for a tunable optical metamaterial surface that is both solid state and can be switched using an applied
voltage.

89

WAVE
WAVE UNDERGRADUATE
RESEARCH FELLOWSHIPS

W
A
V
E

Search for Hidden Broken Symmetries in Strongly Correlated Electron Systems Using Rotational
Anisotropy Optical Reflectivity Measurements
Carina Belvin
Mentors: David Hsieh and Liuyan Zhao
The high-temperature superconductor YBa2Cu3O6+x (YBCO) was first discovered over two decades ago, yet the
pseudogap region of its phase diagram is still not well understood. In fact, a microscopic understanding of the
pseudogap region is lacking for cuprates in general. Two sets of theories of the pseudogap region have been
developed: one set proposes that the pseudogap is a precursor phase to superconductivity, while the other argues
that it is a distinct phase that competes with the superconducting phase. In this paper, we examined the electronic
symmetries of two doping levels of YBCO, 1/8 doped and optimal doped, across the pseudogap transition by
perform- ing nonlinear optical rotational anisotropy measurements.
Sputter Deposition of Ga2O3 and Zn(O,S) on Cu2O for Earth Abundant Photovoltaics
Raymond Blackwell
Mentors: Harry Atwater, Yulia Tolstova, and Stefan Ohlemchenko
Cu2O is a promising material for photovoltaics because it has earth abundant constituent elements, a suitable band
gap, and the potential to be processed at a low cost. As a direct band gap material, it is also possible to absorb
light in a much thinner region. The large discrepancy between theoretical efficiencies and the current record
efficiencies are because of an inability to dope the material, the materials low chemical stability, and the lack of a
satisfactory heterojunction partner. Heterojunction devices are of particular interest because multiple
semiconductors can be used to absorb a larger part of the solar spectrum. Ga2O3 is a possible heterojunction
partner because it has a similar conduction band, thus minimizing the conduction band offset. Cu2O was grown via
thermal oxidization of Cu foils and Ga2O3 was deposited via sputter deposition. Prior to the deposition of Ga2O3,
some of the devices were doped with NaI, one of the few materials shown to dope Cu2O. At the time of writing all
devices have low open-circuit voltages and negligible short-circuit currents, but devices made from NaI doped Cu2O
had especially poor performance. This is likely due to the presence of CuO, meaning further surface optimization
must be done.
Detecting Bias in a Self-Organizing Map of Galaxy Photometry Data
Zachary R. Claytor
Mentors: Peter Capak and Dan Masters
WFIRST will measure distances to billions of galaxies using photometry, but a spectroscopic calibration must be
performed to acquire the necessary precision. In order to estimate the minimum spectroscopy WFIRST will need,
tools were developed to analyze a Self-Organizing Map (SOM), a two-dimensional representation of higherdimensional galaxy photometry. Using these tools, Monte Carlo simulations were performed using different regions
of a SOM to see how photometric error affected the layout of the map. While data sets with recorded photometric
error usually had less than 5 pixels of scatter on a 75 x 150 pixel2 map, data with missing errors caused more
scatter. This lays the groundwork for estimating WFIRSTs spectroscopic requirement. Continuation of this work will
build on the existing tools to predict what amount of spectroscopy will decrease the error in the map to the
required precision, while future work should attempt to adjust for missing measurements in the data.
Experimental and DFT Elucidation of the HER Mechanisms With [Cp*Rh(phen)(MeCN)]2+ for Solar
Energy Conversion
Sydney Corona
Mentors: William Goddard and Samantha Johnson
Efficient storage of solar energy is achievable in photoelectrocatalysis where solar energy systems convert sunlight
into chemical bonds of H2 by the hydrogen evolution reaction (HER).13 Before this method can be explored,
knowledge of the HER mechanism must be increased. Thus, electrocatalytic hydrogen evolution with a
homogeneous rhodium catalyst ([Cp*Rh(phen)(MeCN)]2+) is used here to elucidate the detailed account of this
isolated study.4 Density functional theory (DFT) is used to calculate both intermediate and transition state
structures with the B3LYP and M06 functionals along with the lacvp** and 6-311G**++ basis sets.5 Computational
and experimental results are compared for comprehensive analysis, and preliminary results show agreement
between quantum calculations and experiment, also aligning with Grtzel and Klles previously defined
electrochemical pathway.6 The computational 2-electron reduction potentials is -0.89 V vs. ferrocene, where
experiment is expected to agree. In the subsequent step of the HER mechanism, both experimentalists and
theorists independently arrived at the same unprecedented result, where a (Cp*H)RhI complex was calculated as

91

26.4 kcal/mol lower when compared to the predicted RhIII-H complexes. Current work is waiting on further
experimental affirmation, and computational efforts continue to calculate the possible transition states of the H2
evolution step in hopes to reveal the ever-elusive conclusion to this mechanism.

References:
1. Teets, T. S.; Nocera, D. G. Photocatalytic hydrogen production. Chem. Commun. (Camb). 2011, 47 (33), 9268
9274 DOI: 10.1039/c1cc12390d.
2. Vesborg, P. C. K.; Seger, B.; Chorkendorff, I. Recent Development in Hydrogen Evolution Reaction Catalysts and
Their Practical Implementation. J. Phys. Chem. Lett. 2015, 951957 DOI: 10.1021/acs.jpclett.5b00306.
3. Gray, H. B. Powering the planet with solar fuel. Nat. Chem. 2009, 1 (2), 112 DOI: 10.1038/nchem.206.
4. Blakemore, J. D.; Hernandez, E. S.; Sattler, W.; Hunter, B. M.; Henling, L. M.; Brunschwig, B. S.; Gray, H. B.
Pentamethylcyclopentadienyl rhodium complexes. Polyhedron 2014, 84, 1418 DOI:
10.1016/j.poly.2014.05.022.
5. Bochevarov, A.D.; Harder, E.; Hughes, T.F.; Greenwood, J.R.; Braden, D.A.; Philipp, D.M.; Rinaldo, D.; Halls,
M.D.; Zhang, J.; Friesner, R. A. Jaguar: A high-performance quantum chemistry software program with
strengths in life and materials sciences. Int. J. Quantum Chem. 2013, 113 (18), 21102142.
6. Grtzel, M.; Klle, U. Organometallic Rhodium(ir1) Complexes as Catalysts. Angew. Chemie Int. Ed. English
1987, 26 (6), 567570.
Establishment and Analysis of Metabolic Relationships in Co-Cultures of Sulfur Oxidizing and
Reducing Bacteria
Ana Gonzalez
Mentors: Victoria J. Orphan, Derek A. Smith, and Elizabeth Trembath-Reichert
Sulfur-oxidizing bacteria (SOB) and sulfate-reducing bacteria (SRB) have been found closely associated in
environments like microbial mats, hydrothermal vents and marine and freshwater sediment. It has been proposed
that SOB and SRB participate in a closed syntrophic sulfur cycle, but evidence for such has not been found in the
environment. An SOB and SRB co-culture has been established and bioorthogonal noncanonical amino acid tagging
(BONCAT) coupled with fluorescent in-situ hybridization (FISH) will be performed on the culture. Using these
methods I will be able to both taxonomically identify my cultures through FISH and visualize their metabolic
relationships through BONCAT. A greater understanding of the spatial and metabolic interactions between SOB and
SRB will allow researchers to more thoroughly tackle the question of the existence of a closed syntrophic sulfur
cycle.
SERS Substrate for Detection of miR-10b
Nico Mesyngier
Mentors: Axel Scherer and Sameer Walavalkar
MicroRNA-10b (miR-10b) has been shown to strongly correlate with the presence and severity of glioblastoma.
Even when treated, recurrence rates in patients surpasses 90%. Being able to closely monitor miR-10b therefore
would serve as a powerful tool in diagnosing and treating patients in both the short term and the long term. This
work attempts to achieve that goal using a nano-fabricated substrate with gold nano-bulbs for surface-enhanced
Raman spectroscopy (SERS), which is functionalized with cDNA molecules designed to target this specific miRNA.
Utilizing the electric field enhancement between nano-bulbs when excited by a laser, the Raman spectra of the
miRNA can be amplified sufficiently to be detected at low concentrations. RNA strands are differentiated from DNA
using direct identification, relying on peaks unique to RNA. Future work will involve developing a melting curve of
the hybridization between the cDNA target and the miR-10b in order to limit variant strands of RNA from creating
false positives as well as expanding the capabilities of the SERS substrate to other biological molecules.

92

Developing a Comprehensive Numerical Model for Ice Crystal Growth From Vapor Based on the Cellular
Automata Method
Christopher W. Miller
Mentor: Kenneth G. Libbrecht
Given that a well-developed qualitative understanding of the snow crystal morphology diagram has yet to be
realized, studying snow crystal growth can provide significant insight into the inner workings of crystal and pattern
formation and their relation to various factors such as temperature. In addition to this, analyzing ice crystal growth
from vapor serves as a useful case study for learning more about the dynamics of crystal growth from vapor. Here
we report on our progress on developing a comprehensive computer model capable of numerically growing ice
crystals in well-defined conditions via a numerical model in MATLAB based on the cellular automata method. By
developing a numerical model capable of doing this, one can make quantitative comparison with experimental
results to further understand the various factors that influence ice crystal growth. Although our model has yet to
be finalized, it is particularly useful for performing controlled measurements of crystal growth rates and
morphologies under known conditions for comparison with ice crystals grown in lab. Further improvements on the
model will be made based on the results of numerical fits with our experimental ice crystal growth produced by the
model. Further work will be continued towards this goal.
Gravitational Waves Generated by Binary Black Holes
Joy Moore
Mentors: Christian Ott and Daniel Hemberger
Inspiraling binary black holes are one of the strongest sources of gravitational waves in the universe. The
gravitational waveforms from the late inspiral and merger can be simulated on a supercomputer using numerical
relativity. These waveforms can then be extended to low frequency by hybridizing with a post-Newtonian early
inspiral. These data can be made available to the Laser Interferometer Gravitational-Wave Observatory (LIGO),
which will begin its first science observing run in 2015. Assuming that the numerical simulations are correct, and
that general relativity correctly describes what happens in nature in these extreme gravitational scenarios, these
models of the gravitational waves should improve LIGO's ability to detect gravitational waves.
Design of a Highly Efficient, Cost Effective Anode for Chlorine Evolution and Wastewater Treatment
Daniel Ocasio
Mentors: Michael Hoffmann and Yang Yang
Nearly 20% of the manufacturing expense of the Hoffmann groups self-contained, photovoltaic-powered toilet and
wastewater treatment system is contributed by the cost of the electrode array. This high expenditure is due largely
to the current design of the IrO2 anode used within the wastewater electrolysis cell. In order to reduce this cost, an
alternative anode design was proposed that replaces the costly IrO2 interlayer with a more affordable substance:
RuO2. The anodes were prepared by repetitively brush coating Ru precursor onto pretreated conductive Ti sheets
and annealing until a desired mass loading was achieved. Then an aqueous titanium glycolate complex was
thermally decomposed onto the surface to provide stability. These anodes were then coupled with a stainless steel
cathode and AgCl reference electrode in a single compartment electrolysis cell and electrochemically analyzed with
potentiostatic methods for current efficiency, chlorine evolution rate, and reaction overpotential. Next, each anode
was subjected to an accelerated life test operated at a high current in order to determine the stability. The most
promising samples were then employed in wastewater electrolysis tests to observe their chemical oxygen demand
(COD) and ammonium removal capabilities. Finally, a mechanistic study was performed in order to examine the
pathway by which wastewater treatment is achieved. Overall, the best performing electrode design was comprised
of 1.56 106 /2 RuO2 and 3.76 106 /2 TiO2. The results indicate that the cost-effective RuO2 electrode
is even more efficient than the previous IrO2 design in terms of chlorine evolution and COD removal. However, no
conclusion can yet be made on the stability of the new design until further testing is performed.
Theoretical Study of the Separation of Water Spin Isomers by Selective Adsorption
Jorge L. Rosa
Mentor: Daniel P. Weitekamp
A long-standing controversy is the reported selective physisorption of the para spin isomer of H2O to various
surfaces at room temperature. The theoretical issue is whether the correlation of nuclear spin degrees of freedom
with specific rotational states by the symmetrization postulate can lead to energetic differences in surface or
transition states that are significantly larger than the well-known rotational splittings of the gas phase. Quantifying
how ortho and para water interact with the surface may lead to reliable methods for separation of the spin isomers.
In our study we examine the dependence of the energy and orientation of states of the two spin isomers on
potentials which model the approach to the surface. The computationally inexpensive asymmetric rotor model is
used to calculate energies of the hindered H2O rotors. The talk will outline the symmetry considerations and
numerical computations which underlie this task for a series of increasingly realistic potential energies and coupling
to translational degrees of freedom. Partition functions of the spin isomers in the adsorbed and gas phases are
calculated to evaluate the effectiveness of the model potentials for equilibrium separations.

93

Using Gaussian Process Regression to Infer the Protein Fitness Landscape: A Case Study of
Aziridination Optimization in Cytochrome P450
Raul Sun Han
Mentors: Frances Arnold and Kevin Yang
The advent of protein engineering has provided the means to circumvent natural evolution and directly modify
proteins to generate variants with enhanced or novel functionality. Computational models are vital tools in
traversing the oftentimes vast fitness landscape of proteins and elucidating the relationship between protein
sequence and function. The prospect of optimizing the predictive power of such models is essential for identifying
sequences with optimum functionality, resulting in a more pointed approach to engineering beneficial proteins. To
this end, we train a Gaussian process regression model on experimental data to predict the relative fitness of
protein variants based on the covariance of their structures to those of proteins in the training set. We apply this
technique to a non-contiguous recombinant library of Cytochrome P450s, a superfamily of hemoproteins, to
optimize intermolecular aziridination, a challenging and synthetically useful non-natural reaction.
Exploring Chondroitin Sulfate Protein Interactions
Kimberly Teehera
Mentors: Linda Hsieh-Wilson and Greg Miller
Chondroitin sulfate (CS), a differentially sulfated glycosaminoglycan, serves important roles in the nervous system.
CS is upregulated following injury and inhibits axon regeneration. In particular, the CS-E sulfation motif is a potent
inhibitor of neuronal growth and axon regeneration. CS mediates this inhibitory effect by interacting with various
proteins and modulating intracellular signaling pathways. Known CS interacting proteins include semaphorin 3A
(Sema3A), neuropilin 1 (Nrp1), recticulon-4 receptor (NgR1), and reticulon-4 receptor-like 1 (NgR3). To identify
CS binding sites on these proteins, mutants lacking key basic residues will be generated and interactions with CS-E
polysaccharides will be tested in a binding assay. Currently, wild type (WT) DNA constructs for Sema3A, Nrp1,
NgR1, and NgR3, as well as mutant constructs for Sema3A and Nrp1, have been generated. WT and mutant
proteins will be expressed in HEK293 cells and purified from conditioned media. To test for binding, the protein of
interest will be immobilized to a substratum followed by incubation with biotinylated CS-E. Bound CS-E will be
detected using a streptavidin-horseradish peroxidase conjugate. The identification of CS-E binding sites will further
our understanding of the mechanisms by which CS modulates intracellular signaling pathways through interactions
with soluble protein ligands and cell surface receptors.
HER Mechanism of Rhodium Complexes
Walther Villatoro
Mentors: William A. Goddard, III, and Samantha Johnson
The photosynthetic harvest of hydrogen via hydrogen evolution (HER) is pivotal for powering the planet because it
allows energy to be stored as chemical energy in H2 bonds. In order to be able to store this energy, we must first
gain a deeper understanding of HER. To this aim we elucidate the mechanism of hydrogen evolving rhodium
complexes through computational methods. Quantum mechanical calculations using density functional theory (DFT)
were done on rhodium complexes using B3LYP and MO6 functionals with lacvp** and 6-311G**++ basis sets,
respectively, to elucidate the mechanism of HER by a (Cp*)Rh(bpy)(MeCN) [Cp* = pentamethylcyclopentadienel,
bpy = bipyridine] complex. We determined that this complex first undergoes a two electron reduction before
undergoing protonation. Preliminary computational results show -1.15 V vs Ferrocene (Fc) for a two electron
reduction potential of [(Cp*)Rh(bpy)(MeCN)] which agrees with an experimental result of -1.05 V vs Fc. The first
protonation however, yielded unprecedented results. The experimental group saw a geometry that we calculated in
previous geometry optimizations on a rhodium (V) hydride. In this geometry the proton that was originally thought
to be attached to the metal center relocated to a carbon on the Cp* above the MeCN ligand. The free energy values
calculated for the Rh(I)(Cp*H) was 0.1 kcal/mol which is much more favorable than a 32.5 kcal/mol value
calculated for the Rh(III)(H)(Cp*). Future work involves finding transition states that lead to the actual evolution of
hydrogen.
The Effect of Ablation of Lin28 and let-7 on R-848-Mediated B Cell Responses
Chittampalli (Yasha) Yashaswini
Mentors: David Baltimore and Shuai Jiang
let-7 is a highly conserved microRNA with a variety of functions, but is most well-known for its tumor suppressor
function. Lin28 is a protein that is involved in development, tumorigenesis and metabolism, and has a doublenegative feedback loop with let-7. The objective of this project is to investigate the reaction of splenic B cells from
let-7 knockout mice and Lin28 knockout mice to R-848, a drug with potent antiviral and antitumor activities.
Through R-848 treatment of splenic B cells from let-7 knockout mice and Lin28 knockout mice, and analysis with
FACS and ELISA, we have observed that absence of Lin28a and Lin28b results in decreased R-848-mediated B cell
activation, and that absence of let-7b and let-7d results in increased B cell activation. We have also observed that

94

Lin28b knockouts have decreased antibody secretion, and let-7b and let-7d knockouts have increased antibody
secretion. These results suggest that Lin28 and let-7 play roles in R-848-mediated B cell response, and that let-7b,
let-7d, and Lin28b are involved in antibody diversification via class switch recombination. Suggestions for the
continuation of this project include conducting qPCR and Western Blots to identify which class switching genes and
proteins are regulated by let-7 and Lin28.

95

AMGEN
AMGEN SCHOLARS PROGRAM

A
M
G
E
N

The Claustrum and Insula in Mouse: A High Resolution Diffusion Imaging Study
Hriday Bhambhvani
Mentor: John Allman
The agranular insular area and claustrum are juxtaposed in the mouse, though it is unknown whether the
claustrum has closer affinity for the cortex or striatum. Given this, we compared the connections of the two
structures in mouse brain using high angular resolution diffusion imaging (HARDI, with 72 directions) and
probabilistic fiber tractography. This method allows for the connection of every voxel to be mapped and is nondestructive, whereas conventional tract tracer studies are limited by certain injection sites. Data is forthcoming.
Structural and Functional Analysis of Kap104-Mediated Nuclear Import of Ribosomal Protein L4
Emma Goguen
Mentors: Andr Hoelz and Ferdinand Huber
Currently, little is known about the hierarchical assembly of the 60S and 40S pre-ribosomal subunits within the
nucleus. How the numerous required nucleocytoplasmic transport events occur in a timely coordinated manner still
remains poorly understood. However, the nuclear import factor Kap104 has been identified to facilitate nuclear
import of several ribosomal proteins (RPs), including RpL4, RpL5 and RpL11. Recently, an assembly chaperone for
RpL4 (Acl4) has been shown to bind an internal RpL4 loop and is, after nuclear import, dissociated by interactions
with the surface of the large pre-ribosomal subunit. The aim of this project is to further understand the architecture
of the trimeric import complex and specifically the interactions of Acl4RpL4 with Kap104 that are essential for
complex formation. Therefore, a combined structural and biochemical approach will be used to elucidate the
molecular details of the import of RpL4 into the nucleus. Ultimately, the results of these experiments will shed light
on the nuclear import mechanisms of ribosomal proteins and will provide valuable insight into ribosome biogenesis.
Extending a C-H Functionalization Strategy to Borylation and Phosphination
Gregory C. Han
Mentors: Brian M. Stoltz and Wen-Bo (Boger) Liu
The common, inexpensive potassium tert-butoxide has been shown to catalyze the silylation of heteroaromatic
compounds. In this project, we tested this novel catalytic method with boranes and phosphines, two groups of
compounds which, like silanes, have had their radical species studied in detail. We used 1-methylindole as the
model heteraromatic compound, given its efficacy in the proven silylation chemistry. We have determined that
boranes and phosphorus(III) compounds are unlikely to work with this method, and have instead focused our
efforts on phosphorus(V) compounds.
Molecular Pathways of Bcl11a and Bcl11b in T-Cell Commitment
Ashley Hope
Mentors: Ellen V. Rothenberg and Hiroyuki Hosokawa
The development of T-cells from multipotent progenitors is highly regulated by a complex network of transcription
factors, including Bcl11a and Bcl11b. Cells which do not properly undergo the transition to become committed cells
will continue to proliferate above normal levels and become malignant. In the step-wise progression of T-cell
development, Bcl11a is expressed in early stages when the progenitors are still able to give rise to multiple blood
cell lineages, and is inactivated during commitment, while Bcl11b is activated during commitment, and is highly
expressed during the T-cell specific stages of development. We seek to understand how the transcriptional changes
Bcl11a and Bcl11b make during transition cause the cell to relinquish all other potential to become a T-cell. Using
cDNA from DN3 cell lines infected with retroviruses containing Bcl11a and Bcl11b, we are running RT-qPCR analysis
of a series of over fifty target genes the Rothenberg lab has determined to be significant in T-cell development, to
quantify and analyze the transcriptional effects of Bcl11a and Bcl11b during commitment. Our results reveal not a
simple antagonistic relationship between the opposing factors, but a complex interplay of both cooperative and
conflicting efforts in gene regulation, which draw a new genetic map of commitment.
Developing Mass Spectrometry Assays for Phosphorylation of CamKII
Maria Karelina
Mentors: Mary B. Kennedy, Dylan Bannon, and Leslie Schenker
Phosphorylation plays a major role in cell signaling pathways. Due to the complexity of signaling pathways, a high
throughput method to detect the phosphorylation states of cellular proteins would aid in the understanding of many
biological processes such as Synaptic Plasticity. This study focused on developing assays to identify the
phosphorylation state of CamKII phosphorylation sites with High Performance Liquid Chromatography (HPLC) and
Multiple Reaction Monitoring Mass Spectrometry (MRM-MS). We used Filter Aided Sample Preparation (FASP) to
produce tryptic peptides from hippocampal slices and neural cell cultures. From both types of preparations we were
able to detect de-phosphorylated aCam286, bCamKI287, CamKII305 peptides using HPLC/MS-MRM. We found that

97

the FASP derived from cell cultures had less noise in detection than the FASP derived from slices. The thin layer of
culture is also more susceptible to pharmacological treatments than the thicker slices. This could mean that using
cell cultures for studying the time-course of phosphorylation of proteins may be preferable to use of hippocampal
slices. This can be used to verify these mass spectrometry methods by driving the phosphorylation above baseline
in cultures and detecting a difference in phosphorylation levels between differently treated samples.
Identification of Mitophagy-Regulating Factors: Determining the Role of Rab7 and Other Mitochondrial
Proteins in OXPHOS-Stimulated Mitophagy
Katherine Kim
Mentors: David C. Chan and Rebecca Rojansky
Mitophagy is a specialized form of autophagy that eliminates dysfunctional mitochondria in cells. During mitophagy,
defective mitochondria are selectively targeted for degradation in autolysosomes. It has been found that cells
robustly undergo mitophagy when grown under a specific oxidative culture condition that lacks glucose and
contains acetoacetate, a carbon source that cannot be metabolized by glycolysis. Supplying cells with this oxidative
media increases cellular demand for mitochondrial oxidative phosphorylation (OXPHOS), which results in an overall
higher turnover of mitochondria. To identify regulators of this pathway, we have designed small hairpin RNAs
(shRNAs) targeting proteins that have previously been associated with mitochondrial function. One example is
Rab7, a small GTPase that regulates downstream endocytic trafficking and the biogenesis of lysosomes. Mouse
embryo fibroblasts (MEFs) have been infected with the shRNAs and plated on coverglass slides for microscope
viewing. Through live cell imaging using a confocal microscope, we can analyze the role of our targeted
mitochondrial proteins by searching for any defects or changes in OXPHOS-stimulated mitophagy in the knockdown
cells.
A Study on Assisted Tandem Catalysis: Z-Selective Olefin Metathesis-Oxidative Cyclization Catalyzed by
a Single Ruthenium Species
Daniel Lee
Mentor: Robert H. Grubbs
Grubbs ruthenium catalysts for olefin metathesis are widely used in organic synthesis because of stability,
selectivity, and functional group tolerance. Recent efforts in the Grubbs group have been geared towards using
highly active ruthenium metathesis catalysts exhibiting kinetic Z-selectivity to promote other non-metathesis
chemical transformation and applying them to assisted tandem catalysis. This project focused on investigating
conditions to allow the Ru-metathesis catalyst to serve as an oxidative cyclization catalyst in the presence of an
oxidant in an assisted tandem catalysis protocol. A cyclometalated complex catalyzes first a Z-selective crossmetathesis with a 1,5-diene and a terminal olefin, proceeded by a stereospecific oxidative cyclization. Both steps
are catalyzed by Ru: after cross-metathesis the Ru complex is transformed to an oxidation catalyst upon addition
of NaIO4. 1,5-dienes with sterically bulky substituents were subjected to assisted tandem catalysis. Generating a
complex heterocyclic molecule from relatively simple 1,5-dienes using assisted tandem catalysis is interesting.
Using this new method, a variety of biologically active complex natural products containing tetrahydrofuran diols
can be synthesized.
Biochemical Characterization of the Binding Partners of Sec13
Regina Liu
Mentors: Andr Hoelz and Karsten Thierbach
Eukaryotic cells developed numerous specialized proteinaceous assemblies that perform their cellular processes,
such as intracellular signaling and transport. The small WD40 repeat protein Sec13 was found to be present in at
least three of such assemblies. It is an established member of the COPII machinery involved in anterograde
vesicular transport, a component of the nuclear pore complex (NPC) that serves as the sole mediator of
nucleocytoplasmic transport, and a part of the GATOR complex that is a negative regulator of cellular metabolic
control. The role of Sec13 in these different processes is so far poorly characterized. To better understand the
mechanistic role of Sec13 in its diverse assemblies, we aimed to establish in vitro reconstitution assays using
recombinantly expressed and purified individual proteins. We focused on the expression and purification of Sec16
and Mtc5, known binding partners of Sec13 in the COPII assembly and the GATOR complex, respectively. The
biochemical and structural characterizations of these two proteins and their interaction with Sec13 will potentially
allow us to better understand how Sec13 accommodates binding to different proteins. Ultimately, this would reveal
mechanistic details of Sec13 function in various, essential cellular processes.

98

Investigating and Characterizing Neuropeptide Gene flp-8 on Developmental Fate and Behavior in
Caenorhabditis elegans
Vivian Lu
Mentors: Paul W. Sternberg, James Lee, and Pei Yin Shih
In response to harsh environmental conditions, Caenorhabditis elegans may elect to undergo stress-resistant
developmental arrestcalled the dauer larval stageinstead of transitioning into reproductive adults. Dauer arrest
maximizes C. elegans survival as it allows the organisms to remain metabolically dormant until conditions are
favorable enough to become reproductive. To better understand the regulation of C. elegans developmental
decision-making process, we chose to focus on the neuropeptide gene flp-8, because flp-8 mutants display a
remarkable impairment in their ability to form dauers. The flp-8 overexpression rescue generated rod-like worms
with shrunken intestines that may resemble the constricted intestines of dauers. The genes function was
characterized using Nomarski microscopy, genetic crosses, and behavioral assays. Preliminary results for the
quantification of flp-8 pathology indicate a striking decrease in intestine width to body width ratio compared to wild
type. Furthermore, the cross-progeny of flp-8(+++) hermaphrodites and a fluorescent reporter for the
transcription factor ZC204.12, which is only expressed fully in dauer intestines, have higher GFP intensity than the
fluorescent reporter strain by itself. These results suggest that the flp-8(+++) phenotype may induce intestinal
constriction prematurely after dauer commitment. For future experiments, our goal is to narrow down a possible
mechanism and site of action of flp-8 in order to further understand its function.
Crossmodal Interaction in Audiovisual Steady State Responses Measured via EEG
Kathryn McClain
Mentors: Shinsuke Shimojo and Noelle Stiles
Natural environments change in brightness and loudness at a wide range of rates. Some of these rates are more
relevant to people than others, for example, the rate of the human gait is important to recognizing other humans.
Therefore, it is not surprising that the strength of crossmodal interactions vary across stimuli rates. This project
investigates these crossmodal interactions by measuring the neural response to beeps and flashes at varying rates
via electroencephalogram (EEG). These EEG recordings have shown that the electrical activity tends to sync with
the stimuli and fluctuate at the on/off rate of the stimuli. We tested 3 different modalities (auditory, visual, and
audiovisual) at 4 different rates (3hz, 6hz, 10hz, 15hz). In each modality, different areas of the brain show
maximal synchrony with the stimulus. The audiovisual responses have shown similar activation patterns to the
visual responses, with some areas more strongly synchronized by the audiovisual stimuli than the visual alone,
possibly demonstrating crossmodal integration between audition and vision. Different frequencies within a modality
have also shown differing response patterns, which may indicate a natural responsiveness of some neural
populations to particular frequencies. These results may clarify the neural networks that perform crossmodal
integration of temporal information.
Synthesizing a Bone Mimetic Scaffold Through the Electrochemical Deposition of Mineralized Collagen
Luizetta Navrazhnyh
Mentors: Julia Greer, Ottman Tertuliano, and Alessandor Maggi
Grafts can assist in treating bone disorders by functioning as frameworks for the growth on new bone, providing
cell anchoring sites, mechanical stability and structural guidance. Ideally, the graft would integrate into the natural
remodeling process and be gradually replaces with bone. However, collagen and hydroxyapatite, the dominant
components of natural bone, are not frequently used in grafts since independently collagen is too weak while
hydroxyapatite is too brittle for use in a load bearing structure. The goal of this project is to combine collagen and
hydroxyapatite in a composite structure similar to natural bone and to examine emergent mechanical properties.
Therefore films of varying morphology and composition were synthesized by concurrent electrochemical deposition
of collagen and calcium phosphate. Their chemical composition was analyzed by Raman, FT-IR and Energy
Dispersive X-ray spectroscopy and their morphology was examined by SEM. The films will be deposited into pillar
structures to examine their compressive strength and hardness. An advantage of the electrochemical deposition
method is the ability to deposit material into 3D molds. Thus the synthesized material can subsequently be
deposited into 3D structures, allowing geometric effects to improve mechanical properties.
A Neurocognitive Computational Approach to Understanding Theory of Mind
Isabelle A. Rosenthal
Mentors: Ralph Adolphs and Damian Stanley
Inferring the beliefs and intentions of others (Theory of Mind; ToM) is critical for social cognition. Impaired ToM is a
common characteristic of Autism Spectrum Disorder (ASD). While previous work has implicated certain brain
structures as mediators of ToM processing, little is known about the specific computations they perform or what
goes awry in ASD. We addressed this question using a novel task that permits computational modeling of ToM
learning processes, while retaining the core features of traditional ToM tasks. Participants learned continuously
about the intentions and beliefs of another person (the Agent) and integrated this information to make predictions

99

about the Agents choices. Earlier work showed behavioral differences in control and ASD task performance. Here,
we extend this research by adapting and testing the task to be viable for neuroimaging (MRI). By pairing MRI data
and behavioral data with previously constructed computational models of learning, we hope to gain a better
understanding of the neural correlates of ToM in both controls and ASD.
Identifying Factors Involved in a Mouses Behavioral Response to a Looming Visual Threat
Sarah Sam
Mentors: Markus Meister, Kyu Lee, and Zeynep Turan
Studying behaviors triggered by visual stimuli in the mouse could provide insight into the role of visual processes in
cognition. Previously, the Meister Lab discovered that mice exhibit a defensive behavior in response to a looming
stimulus. An expanding black disc is shown above a mouse in a rectangular arena. This stimulus, which simulates
an approaching aerial predator, elicits one of two behavioral responses: the mouse either flees the area to a nest
or freezes in place. Despite the fact that these are laboratory mice raised in a controlled setting without any
predators, they respond robustly to this stimulus, indicating evolutionary significance in these behaviors. Mice may
be using a combination of apparent velocity of the looming stimulus and cognitive map of its surroundings, among
other factors, to decide between freezing and fleeing. Identifying the factors that govern this decision-making will
help understand the underlying neural circuitry involved in this behavior. We find that fleeing occurs most often
when there is a nest present. Mice prefer to flee when the stimulus is presented less than three feet from the nest.
Finally, we will assess whether the mouse is using a cognitive map of its surroundings to make the decision.
The Predicted Ensemble of 3 Dimensional Structures for Human Olfactory Receptors
Christian Seitz
Mentors: William Goddard, III, and Soo-Kyung Kim
Human olfactory receptors (hORs) are members of Class A G protein-coupled receptors (GPCRs), which is the
largest class of GPCRs; odorants binding to hORs give us our sense of smell. However, the 3 dimensional structure
and binding details of most hORs are not known. The energetically-best 3 dimensional conformations were
predicted for three hORs (OR1A1-4, OR1D2-4, and OR3A1-4) using Monte Carlo sampling to sample ~100 million
structures from ~10 trillion possible conformations. The three angles sampled for each transmembrane helix were
, the helix rotation angle about its own axis; , the tilt angle of the helical axis; and , the helix sweep angle that
specifies the direction of the helical axis tilt. The top 25 structures for each hOR were ranked according to energy,
and three structures from each hOR were selected for ligand docking using criteria including structural diversity,
hydrogen bond energy, and conserved 1-2-7 and 3-6 interactions. Each hOR was docked with odorants that had
been previously determined to exhibit binding to the target hORs. Once odorants were docked to hORs, five cycles
of quench annealing were run. These results detail the key active/inactive conformations in three hORs and the key
binding residues for different odorants.
The Role of hoxb5b in Zebrafish Neural Crest Development and Migration
Joshua Tworig
Mentors: Marianne Bronner and Rosa Uribe
Neural crest cells (NCC) are multipotent migratory stem cells that develop into a variety of cell types. Originating
from the neural tube, they are guided during their migration by molecular and environmental signals that lead to
proper developmental patterning. The transcription factor Hoxb5 is expressed in a subset of migratory NCC, as well
as in their migratory environment. Hoxb5 plays a functional role in NCC migration in amniotes, such as chicken and
mice. Here, we investigate the role of Hoxb5b, a zebrafish orthologue of Hoxb5, in the development of NCC during
zebrafish development. We hypothesize that hoxb5b overexpression leads to aberrant NCC growth and migration.
To test this hypothesis, we injected zebrafish embryos with hoxb5b-vp16 mRNA, encoding Hoxb5b protein with
enhanced transcriptional activation capabilities. Utilizing immunohistochemistry and in situ hybridization in
combination with various imaging techniques, we characterize subsequent NCC patterning at a series of
developmental time points. We find that ectopic Hoxb5b signaling leads to global phenotypic effects including
increased migration and proliferation of NCC. These findings support the hypothesis that Hoxb5 signaling plays an
important conserved role during NCC development and provide relevance for future studies directed toward of NCC
and the role of hoxb5b in their development.
Electrochemistry of the Heme Domain Cytochrome P450 BM3 Protein
Adelfa Amanda Vera Rosado
Mentors: Harry B. Gray, Bryan Hunter, Mike Hill, and Andrew Udit
The cytochrome P450 is a superfamily of proteins that performs regio- and stereospecific oxidation reactions that
offer a green alternative for the biosynthesis of many drugs. The fatty acid hydroxylase CYP102A1 (P450 BM3),
shows the highest catalytic hydroxylation rate, and some mutants have shown higher turnover rate than the wild
type. However, expensive NAD(P)H cofactor is needed to activate the protein, and the peroxide shunt formed by
the dioxygen impedes the completion of the catalytic cycle. The purpose of this project is to make advances in the
development of reversible electron transfer systems for cytochrome P450 BM3 biocatalysis, in order to optimize

100

their synthetic potential for biotechnological applications. We report the synthesis of the P450 BM3 wild type, and
of three mutants. Glassy carbon electrodes were coated with different concentrations of azidoaniline via
electrochemical grafting, using cyclic voltammetry. Subsequent work includes the covalent attachment of a quinone
linker to the electrode, and the succeeding attachment of the P450 BM3 mutants to the quinone to complete the
electrode system. Cyclic voltammetry and X-ray photoelectron spectroscopy will be used to characterize the
system, and the preliminary findings will be discussed.
Mimicking Enzyme Activity With Macrocyclic Peptides
Andrew Wang
Mentors: James R. Heath and Blake Farrow
Enzymes, though highly complex and evolutionarily tailored structures, can be viewed in a reductionist sense as a
scaffold holding several critical residues and reactive cofactors in a reactive, entropically constrained system. It is
possible to envision simpler alternatives to the enzyme scaffold which are nonetheless able to perform the function
of arranging residues in a proper orientation to render them active. One such alternative are conformationally
constrained synthetic macrocyclic peptides, previously used by our group to develop small-molecule antibody
mimics. These peptides also readily allow incorporation of noncanonical amino acids, which may be able to mimic
critical enzymatic cofactors. Using a comprehensive library of 5-amino acid macrocyclic peptides and screening
against a serine hydrolase suicide substrate which only binds irreversibly to enzyme active sites, we have identified
molecules which display activity against a serine hydrolase ABP. Furthermore, incorporating a methyl-thiazolium
noncanonical amino acid into our screen as a thiamine mimic, we have identified macrocyclic peptides which
display activity against a pyruvate decarboxylase suicide substrate. Characterization of these peptides in NMR
reveals significant perturbation of the local chemical environment in active macrocycles. The possibility of
developing small peptide mimics of enzyme activity offers an interesting and novel approach to catalysis.
Determining the Mechanism by Which the Ssa1-Sgt2 Chaperone Network Captures and Selects
Tail Anchored Proteins
Kathleen Yu
Mentors: Shu-ou Shan and Hyun Ju Cho
Accurate delivery of newly synthesized membrane proteins is essential for cell structure and function. Tail anchored
(TA) proteins, which have C-terminal transmembrane domains (TMDs), must be post-translationally inserted into
the endoplasmic reticulum (ER) by chaperone proteins through the GET (Guided Entry of Tail Anchored Proteins)
pathway. As none of the pathways components associate with ribosomes, it is uncertain how TA proteins are
selectively recruited from their site of synthesis and directed to the GET pathway. Sgt2, the first known component
for initial TA substrate capture, contains a tetratricopeptide repeat domain that binds heat shock proteins. Prior
experiments have shown that the heat shock protein Ssa1 (the most abundant yeast Hsp70) rapidly binds Sgt2 in a
reversible manner, suggesting that Ssa1 may serve as a hub that sorts ER-resident proteins to Sgt2. The aim is to
determine the role of the Sgt2-Ssa1 chaperone network in TA protein capture and selection to the GET pathway.
We will fluorescently label a TA protein (Bos1) containing mutations that make it an increasingly poor substrate for
the GET pathway. We will perform real-time and equilibrium titration FRET measurements between the TMD of TA
protein (donor probe) and the C-terminus of Sgt2 (acceptor probe). These TA transfer assays will allow us to
compare the rates and equilibria of Ssa1 chaperone activity for TA variants in order to define the function of Hsp70
in Sgt2-TA substrate binding.
Predicting and Validating the Ensemble of 3-Dimensional Structures for Group V Human Olfactory
Receptors
David Zhou
Mentors: William Goddard, III, and Soo-Kyung Kim
Human olfactory receptors (hORs) are Class A G protein-coupled receptors (GPCRs) that allow humans to detect
odorants and perceive smell. Currently, the odorant binding details and activation mechanisms for each hOR are
unknown because it is difficult to obtain crystal structures for GPCRs. In recent years, structures for hORs in each
of the six clusters, including the receptor OR1G1, have been predicted using the Bihelix and SuperBihelix
computational methods for predicting GPCR structure. These structures were used as templates to predict the
ensemble of 3-dimensional structures for three hORs in the group V cluster (OR2M7-5, OR2J2-5, and OR2C1-5)
using the same methods. These methods use Monte Carlo sampling of the three angles that define the position of
each of the seven transmembrane helices to sample 67,500 structures from approximately 10 trillion possible
unique conformations. The sampled structures were then ranked to find the 25 most energetically favorable
conformations. Using the DarwinDock method, which predicts the position and orientation of a ligand to a protein

101

LIGO
THE LASER INTERFEROMETER
GRAVITATIONAL-WAVE OBSERVATORY

L
I
G
O

Determining the Effect of Acoustic Coupling on Advanced LIGO

Katherine Banowetz
Mentor: Robert Schofield
Advanced LIGO is built to be extremely sensitive to movements of the test mass as small as 10-20 m/Hz, which
allows many signals other than gravitational waves to be detected by the system. Pressure created by external
sound can alter the differential arm movement measurement by creating Doppler shifts, intensity fluctuations, and
scattering in the laser beam. To determine the areas affected by external sound, we inject acoustic noise in the
laser and vacuum equipment area. On a smaller scale, vibrating a horizontal access module or beam splitter
chamber with a shaker tests the effect of sound on single chambers. To calculate the scale at which these
vibrations affect the differential arm movement signal as well as the effect of other environmental injections, I
created a Python program. This program analyzes ambient background noise signals as well as injections with
coupling functions and outputs a file with the estimated differential arm movement effect for each calculated
frequency. By calculating the effect of acoustic coupling and other environmental signals with this program, the
calculation process will be streamlined and calculation error will be reduced.
Tracking Spectral Noise Lines in Advanced LIGO Data
Gillian Dora Beltz-Mohrmann
Mentors: Alan Weinstein and Jonah Kanner
The Advanced LIGO detectors are expected to make gravitational wave observations possible within the next few
years. However, sharp spectral noise lines continue to obscure the data, and it is unknown if or how these lines
wander over time. Therefore, we are developing a method that will track the frequencies of the various noise
sources which appear in our data. Using Python for scripting, we utilize various signal processing techniques to
extract information about the frequencies present in our time series. We then heterodyne to examine how a given
spectral line wanders in frequency over time. In the future, this method will be automated to constantly examine
new data in quasi-real time, providing beneficial insight for improving the quality of the data and the sensitivity to
gravitational waves from spinning neutron stars and other astrophysical sources.
Detector Characterization of the LIGO 40m Prototype Interferometer
Eve Chase
Mentors: Koji Arai and Maximiliano Isi
The Advanced Laser Interferometer Gravitational-Wave Observatory (aLIGO) detectors will come online in the fall
of 2015 with the goal of providing the first direct detection of gravitational waves. Detector characterization
techniques, focused on the reduction of noise, are imperative to analyze and suppress noise to provide sufficient
sensitivity for the detection of gravitational waves. Caltech houses the LIGO 40 meter prototype interferometer, a
copy of the LIGO interferometers, providing the perfect playground to test and design novel detector
characterization techniques for later implementation at the main sites. While summary pages, websites showing
the state of the detector in real-time, are used to monitor the main detectors, the 40 meter prototype lacks many
of these low-latency monitoring features. We incorporate new features into the 40 meter prototype summary pages
in the hope of benefiting the entire LIGO community by providing real-time access to detector monitoring tools.
Additionally, we develop basic tools to distinguish between gravitational wave signals and non-Gaussian noise
sources, which are often misidentified as false-positive gravitational wave signals.
Sensing and Control of Suspended Optic Breadboard in Crackle2 Experiment
Sai Kanth Dacha
Mentors: Gabriele Vajente and Rana Adhikari
Crackling noise arises when a system responds to external conditions through discrete, impulsive releases of
energy. Various locations in the Advanced LIGO setup have been identified over the years as suspected to produce
crackling noise. Of specific interest is the crackling noise arising in maraging steel blades that are used to suspend
end mirrors of the Michelson Interferometer in a quadruple pendulum stage, where vertical displacement noise due
to crackle events couples to horizontal displacement noise along the arm of the interferometer.
In order to study this noise experimentally, a setup was designed a couple of years ago: a Michelson
Interferometer configuration to measure vertical displacement noise at the tip of the maraging steel blade. Our
work described in this paper focuses on establishing a feedback damping control loop around the suspended opticsbreadboard that houses the interferometer. We describe the whole process of control system design, starting from
sensing to actuation and control. We also discuss certain miscellaneous things we have worked on to improve
speed and efficiency for further rounds of implementing the control system, upon any possible modifications to the
setup. We conclude by talking about the improvement the new control system has brought about in damping
motion of the breadboard at the suspension resonant frequencies.

103

A Three Dimensional FFT Based Search Scheme for Gravitational Waves From Binary Neutron
Star Systems
Deanna Emery
Mentor: Richard Gustafson
Gravitational waves (GW) from binary neutron stars are phase modulated due to a Doppler shift caused by the
orbital motion of the stars. This phase modulated signal will have an oscillating frequency. Finding the carrier
frequency of a phase-modulated signal after applying the Fourier transform is non-trivial. This is due to the
modulation factor of the signal, which induces sidebands that vary in magnitude relative to the carrier frequency
depending on the size of the modulation index (the extent of the Doppler shift). A GW signal can be extracted from
noise using a data search over three parameters: the carrier frequency (ranging from 10-1500 Hz), the modulation
frequency (about 0.0001-1 Hz), and the modulation index/amplitude (0-1000). Key strategies for this search
include isolating the modulation factor from the signal and approximating it using the Jacobi-Anger expansion,
working in the frequency domain (taking the Fourier Transform only once in the initial step), and utilizing the
Convolution Theorem. The program for this search has been able to determine all three parameters from simulated
data both with and without noise, though with some limitations that depend on the number of terms used in the
Jacobi-Anger expansion and the amount of noise.
Testing the Strong Field Dynamics of General Relativity Using Compact Binary Systems
Carl Fields
Mentors: Tjonnie Li and Alan Weinstein
Einstein's General Theory of Relativity (GR) has been well tested in the weak field regime over the past century.
However, such tests have not been carried out in the highly dynamical and inherently non-linear strong field
regime. Recent advancements in ground based gravitational wave detectors, (e.g., Advanced LIGO, VIRGO), will
allow us probe this regime of general relativity by investigating gravitational waves produced by astrophysical
systems with strong gravitational fields such as compact binary coalescences. While current search techniques
utilize standard GR waveforms to identify weak GW signals in the presence of noisy data, alternative theories of
gravity predict signals that differ significantly from GR. We investigate our ability to find non-GR effects in detected
waveforms of an astrophysical source in an alternative theory of gravity by introducing an arbitrary parameter,
nGR, to modify standard GR waveform features, such as ringdown frequency, merger frequency, and amplitude.
We then perform statistical methods such as matched filtering and bayesian inference to quantify how well future
detectors will be able to distinguish between the gravitational waveforms in the event that GR is not the complete
theory of gravity.
Assessing the Merit of a Space-Based Gravitational Wave Antenna for Cosmography and
Black Hole Spectroscopy
Jared Goldberg
Mentors: Rana Adhikari, Curt Cutler, and Tom Callister
Physicists are creating a network of gravitational wave (GW) detectors on Earth, which would be capable of
detecting signals from 10-103 Hz from low-mass compact binary systems. In order to observe more massive
binaries, which coalesce at frequencies inaccessible to ground-based detectors due to seismic and atmospheric
noise, plans have also been made to create space-based detectors. However, the most serious of these proposals,
LISA, is not scheduled for launch until 2034 due to its extreme cost. Therefore, recent efforts have aimed at
designing lower-cost detectors that could span a similar frequency band. This project aims to evaluate the scientific
merit of one such proposed space-based detector. The proposed detector is sensitive to GW signals in the range of
10-3-102 Hz, overlapping with the earth-based detectors. We calculate the detectors best-case signal-to-noise ratio
as a function of a sources location and orientation, and using the Fisher Matrix formalism, we calculate how
accurately the detector could locate sources in the sky. We find that, while the detector can strongly detect GWs
out to cosmological distances, its noise curve is too high to accurately pinpoint its sources. However, upon
synthesizing this detectors data with data from earth-based detectors, the detector may still be able to locate
interesting GW sources.
Excess Current Noise in High Current Photodetection
Arjun Iyer
Mentors: Eric Gustafson, Rana Adhikari, and Zach Korth
Advanced LIGO (aLIGO) in its quest to detect gravitational waves uses a Dual Recycled Michelson interferometer to
read the relative phase difference in its two arms induced by the passage of a gravitational wave. These phase
measurements ultimately boil down to a current measurement from a photodiode. This very process of
measurement and the hence the final readout will be corrupted by various noises sources. The noise we concern
ourselves in this project is the excess noise in photodiodes, which will be observed in addition to the ever present
thermal noise of the photoreceiver. We measure this excess noise using the heterodyne demodulation technique.

104

Additionally, we explore a variety of excess noise analysis techniques like 0-90 subtraction method, crosscorrelation method etc. to measure and characterize the excess noise in photodiodes and to identify the excess
noise mechanism in them.
Experimental Study of Crackling Noise as Micro-Mechanics of Flow
Kla Karava
Mentors: Xiaoyue Ni and Rana Adhikari
In crystalline materials, dislocation dynamics can generate strain noise upon being applied rapidly-changing,
oscillatory stress in the elastic regime. This so-called crackling noise may arise inside the maraging steel blades of
the quadruple pendulum which serves as the suspension system of the Advanced LIGO. The study focuses on the
nature and origin of crackling noise in metals. Monocrystalline copper nanopillars are used as model system for this
purpose. Sinusoidal load perturbation to quasi-static compressive loading was applied to copper nanopillars with
different system sizes, and the loss and storage moduli were measured as function of static stress. Amorphous
fused silica nanopillars were also studied as to eliminate non-dislocation-dynamics effects. It was found that the
loss moduli, which characterize the energy dissipation mechanism of the system, are not constant throughout the
static, elastic loading in copper; we also observed size dependence of this anomalous dynamic mechanical
behavior. Our results confirm a deviation from the perfectly elastic behavior which can further indicate intermittent
dislocation dynamics as the source of crackling noise.
Prototyping a Tilt-Free Seismometer
Megan Kelley
Mentors: Kate Dooley and Rana Adhikari
The Laser Interferometer Gravitational-wave Observatory (LIGO) aims to be the first experiment to directly detect
gravitational waves. It consists of a Michelson interferometer constructed to measure the changing distance
between two test masses on the order of 10-21 meters. This level of precision requires excellent noise reduction,
including passive and active isolation from seismic events. LIGOs current seismometers function well at high
(>100Hz) frequencies, but become less effective at low (<10Hz) frequencies. At these lower frequencies, tilt of the
ground begins to contaminate measurements of the translation of the ground. This summers project consists of
prototyping a seismometer that is mechanically insensitive to tilt in the frequency ranges in which LIGO is
sensitive. The first component of the project is the construction of a Michelson interferometer to measure the
motion of the seismometer when subjected to ground motion, and the second component is the construction of a
thermal enclosure and associated temperature control system in order to isolate the system from large
temperature fluctuations.
Cosmography and Black Hole Spectroscopy by Coherent Synthesis of the Terrestrial and Space GW
Antennae Network: Orbit Optimization
Elvira Kinzina
Mentors: Rana Adhikari, Tom Callister, and Curt Cutler
Gravitational waves (GWs), oscillations of spacetime that propagate at the speed of light, may soon become a very
helpful tool for exploring the universe from an absolutely novel point of view. All gravitational wave detectors
existing at the moment are ground-based and therefore are limited by seismic noise at low frequencies. Thus new
concepts of space detectors have been presented. One possibility is the Laser Interferometer Space Antenna
(LISA), a constellation of three satellites which forms an equilateral triangle with million kilometer long arms. But
due to its large cost it will likely not be launched until the mid-2030s. Thus other new simpler and cheaper LISAlike missions have to be proposed. One alternative to LISA is a smaller mission, using 100 kilometer arms and
fewer lasers. In order to measure gravitational waves, this satellite formation must remain stable over a several
year period. Here, we determine the most stable orbital configuration by determining the optimal parameters of the
satellites such as initial positions of the spacecraft and their velocities, the semi-major axis of a desired orbit,
inclination and arm length.
Multi-Carrier Optimization of Future Laser Gravitational-Wave Detectors
Eugene Knyazev
Mentor: Yanbei Chen
Sensitivity of second generation of GW detectors will be quantum noise limited. In this project we assume two main
methods to improve the sensitivity: the first approach comprise the use of double optical spring in order to
increase the dynamical response of test objects to gravitational wave. The effect is called "negative optical inertia"
and it allows to reduce total quantum noise in low frequency band. The second method is based on use of double
antisymmetric carriers in order to reach high frequency improvement. We propose to combine two method
mentioned above in the xylophone configuration to achieve better broadband sensitivity.

105

Searching for Gravitational Waves From the Coalescence of High Mass Black Hole Binaries
Ka Tung Lau
Mentors: Tjonnie Li, Alan Weinstein, Surabhi Sachdev, and Kent Blackburn
The coalescence of binary black holes is a promising class of sources of gravitational waves which can be detected
by ground-based gravitational-wave detectors. The frequencies of gravitational waves generated by the
coalescence of stellar-mass black holes lie in the advanced LIGO (aLIGO) frequency band. aLIGO uses a search
pipeline called GstLAL to search for coalescence signals from the detector output. This search pipeline uses
matched filtering to compute the signal-to-noise ratio (SNR) and x value of the detector signal. The maxima in the
SNR time series which have SNRs higher than a threshold are known as triggers. The challenge is to discriminate
triggers induced by gravitational waves from those induced by noise based on the output of the matched filter. In
this project, we investigate the use of machine learning to achieve this goal. Real detector data and simulated
signals are used as input to the classifier for training and evaluation. The ranking statistic calculated from machine
learning is then compared to the likelihood ratio which is currently used by GstLAL. The use of machine learning
allows us to be more flexible to include additional information such as data quality from auxiliary channels.
A Simulation Study of Lock Acquisition in the Dual-Recycled Michelson Interferometer
Qing Li
Mentor: Kiwamu Izumi
Lock acquisition is the process of bringing each optic in the interferometer to its operating point. In advanced LIGO,
the implementation of the Dual-Recycled Michelson Interferometer (DRMI) adds difficulties to lock acquisition due
to the fact that individual cavity length signals in DRMI are highly non-linear and cross-coupled. In addition, several
parameters like the choice of triggering signals, triggering threshold values, and filter shapes that have significant
impacts on DRMI lock acquisition have been determined empirically and not well understood. In this project, we
develop a numerical simulation of advanced LIGO using the End-To-End module under realistic constrains while
acquiring the flexibility of modifying the control system to study the locking behavior under different
conditions.Using the simulation, we investigate the influence of possible parameters and search for the optimum
method of DRMI lock acquisition.
Searching for Gravitational Waves From the Coalescence of High Mass Black Hole Binaries
Johnathon Lowery
Mentors: Surabhi Sachdev, Tjonnie Li, Kent Blackburn, and Alan Weinstein
Solutions to Einstein's field equations predict gravitational waves: disturbances in space-time that propagate at the
speed of light. Detecting gravitational waves is challenging because the signals are very weak and so a very
sensitive instrument is required. The Laser Interferometric Gravitational-Wave Observatory (LIGO) is a pair of
detectors that search for these disturbances by looking for small length changes caused by passing gravitational
waves. Output from the detectors is analyzed using an advanced form of matched filtering which helps to pull out
small signals from a background of noise. One promising class of sources of gravitational waves is binary black
holes. This progress report details steps taken toward optimizing the analysis pipelines for Advanced LIGO in the
context of binary black hole detection. In addition, I discuss a novel method for the generation of stochastic
template banks which is significantly faster than the traditional method.
Feedforward Seismic Noise Cancellation at the 40M Interferometer
Ignacio Magaa Hernandez
Mentors: Eric Quintero, Koji Arai, and Rana Adhikari
The Advanced LIGO detectors are expected to reach design sensitivity within the next four years. Seismic motion
provides a low frequency limit to the current LIGO detectors design noise curves. At the 40m Prototype
Interferometer, we will test new noise cancellation techniques to mitigate seismic noise. We will implement IIR
Wiener filtering techniques to the mode cleaner and arms cavities. We will also determine the self noise of the
seismic sensors, (accelerometers and seismometers) employed in the interferometer. With this knowledge, we will
explore ways in which this sensors can be better arranged around the 40m interferometer in order to best mitigate
seismic noise. The technologies developed will be applied to the Advanced LIGO detectors in order to improve low
frequency sensitivity to astrophysical events, and progress made on seismic noise cancellation will be used to
reduce other noise sources, particularly Newtonian gradient noise in future detectors.

106

Parametrization of the Effect of Weak Interactions on Production of Heavy Elements in Binary

Neutron Star Mergers
Sandra Ning
Mentors: Hannalore Gerling-Dunsmore and Luke Roberts
Recent research has shown that material ejected during binary neutron star mergers (BNSMs) is a likely source of
the universe's heavy nuclei created via rapid neutron capture, or the r-process. We investigate the previously
neglected effect of positron, electron, electron neutrino and antineutrino capture reactions on the r-process in
BNSM ejecta. To do this, we modify the nuclear reaction network SkyNet's pre-existing r-process simulation code
to include the aforementioned interactions. Then, we parametrize the neutrino luminosities and mean energies and
study their effect on the final abundance of elements in BNSM ejecta, as well as the effects other parameters have
on the weak interaction rates. We then compare our results to the observed solar r-process abundances. Results
show that neutrino luminosities at or above 10^53 ergs/s may significantly limit the production of heavy elements
of mass number >120. We also find the entropy of the BNSM ejecta plays a significant role in determining the
relative importance of electron/positron capture to neutrino captures. Further research may investigate more
accurate modeling of the properties of the neutrino field and the effect of neutrino interactions with the heavy
elements in the ejected material.
Quantization Noise Analysis in Advanced LIGO Digital Control Systems
Ayush Pandey
Mentors: Christopher Wipf and Jameson G. Rollins
Advanced LIGO implements hundreds of control loops using digital signal processing techniques. These controllers
are known to inject noise due to round-off error as signals are IIR filtered, and converted to and from the digital
domain. In this project, the origins of this quantization noise, and its analysis have been investigated for the digital
controller implemented at the Advanced LIGO sites. The quantization noise analysis is done for all its sources viz.
the digital filters, Analog to Digital Converters(ADCs) and Digital to Analog Converters (DACs). The quantization
noise in the filters depend on their structure which has been thoroughly analyzed in this study. The report would
also summarize the results of quantization noise analysis in the DAC code and noise shaping implementation so
that the quantization noise level is mitigated in the frequency band of interest to gravitational wave detection.
Adaptive Feedforward Seismic Noise Cancellation at the 40m Interferometer
Jessica Pea
Mentors: Eric Quintero, Koji Arai, and Rana Adhikari
The interferometers used by Laser Interferometer Gravitational Wave Observatory (LIGO) are subject to a
multitude of noise sources. These noise sources, if not properly accounted for, will eliminate the possibility of
gravitational wave detection. Seismic noise and the self-noise of devices used are two main noise sources that can
be filtered out using a variety of physical and mathematical filters. The goal of this project was to develop
mathematical filters to reduce self-noise of our seismometers and accelerometers, both of which are used to
measure seismic noise. I used both FIR and IIR Wiener filters, in combination with pre-filtering, to account for
these noise sources. The main pre-filter explored was an elliptic bandpass filter, focusing on the region between
3 Hz and 20 Hz, where we expect the instruments self-noise interferes most with the interferometer signal.
Combining such a pre-filter with a Wiener filtered out a significant amount of instrument self-noise.
Cavity Alignment Using Fringe Scanning
Laura Paulina Sinkunaite
Mentors: Michael Landry and Keita Kawabe
LIGO employs two 4-km long FabryProt arm cavities, which need to be aligned in order for an interferometer to
be locked on a TEM00 mode. Once the cavity is locked, alignment signals can be derived from wave-front sensors
which measure the TEM01 mode content. However, the alignment state is not always good enough for locking on
TEM00. Even when this is the case, the alignment can be evaluated using a free swinging cavity, that shows flashes
when higher-order modes become resonant. By moving test masses, small changes are made to the mirror
orientation, and hence the TEM00 mode can be optimised iteratively. Currently, this is a manual procedure, and
thus it is very time-consuming. Therefore, this project is aimed to study another possible way to lock the cavity on
the TEM00 mode. Misalignment information can also be extracted from the power of the higher-order modes
transmitted through the cavity. This report presents an algorithm for this alternative and faster way to derive the
alignment state of the arm cavities.

107

Homodyne Detector Characterization

John Stearns
Mentor: Ryan DeRosa
Currently, Advanced LIGO makes use of a DC readout scheme, a type of heterodyne detection in which the light
containing the gravitational wave signal is sensed against a local oscillator field generated by a static differential
arm length. An alternative scheme is balanced homodyne detection, in which an external local oscillator is
combined with the signal beam on a beamsplitter at the dark port. Both outputs are detected and subtracted to
yield the gravitational wave readout. This scheme has several advantages, including common mode noise
cancellation and facilitating the measurement of sub-quantum noise provided by squeezed light. This work is an
evaluation of the difficulties associated with operating such a detector in air and achieving a high degree of noise
isolation.
Scattering Noise Studies at the LIGO Livingston Observatory via Injection Techniques
Stephen Trembath-Reichert
Mentors: Anamaria Effler and Valery Frolov
As the latest generation of LIGO gravitational wave detectors approaches its first science run in the coming
months, it is becoming increasingly important to specifically identify and target sources of noise, especially
considering the large improvements expected from aLIGO's strain sensitivity levels. It is the purpose of this project
to search for the sources of scattering noise at the LIGO Livingston Observatory in Livingston, LA. We are
concentrating on finding acoustic couplings, as well as identifying the mechanisms where possible. Our main
methods of searching for these sources involve applying injections from various sources on specific parts of the
interferometer and observing the resulting effects on the signals detected from other parts of the interferometer.
In assessing the effects of these injections, we look at the response in the differential arm signal and then use
auxiliary channels to quantify the mechanisms responsible and their coupling parameters as much as possible.
Using these techniques, we have determined the sources of several noise sources due to both linear and non-linear
noise-creating mechanisms within the LIGO sensitive band. These tests will inform future work on reducing scatter
coupling and noise sources.
Inferring the Neutron-Star Equation of State Using Gravitational Waves
Samantha Usman
Mentors: Tjonnie Li and Alan Weinstein
We study how the Laser Interferometer Gravitational-Wave Observatory (LIGO) can be used to learn more about
the nuclear equation-of state. The equation-of-state for nuclear matter describes how matter behaves at high
densities. Since we cannot recreate such high densities on Earth, we turn to astrophysical sources to learn more
about their structure. Gravitational waves can give us insight into the nature of this type of matter. First, we
describe the current state of research on the subject and the motivation behind this project. We then show how
gravitational-wave detections from binary neutron-star systems are sensitive to variations in tidal deformabilities.
This information about neutron stars' tidal deformability will constrain the nuclear equation-of-state. To estimate
this parameter, we use an analysis pipeline which we describe here. This estimation will allow us to quantify the
pipeline's precision as a function of gravitational-wave detections. The deformability will give us new information
about the nuclear equation-of-state and will rule out unlikely models of the matter. Finally, we compare the
pipeline's performance on three possible nuclear equation-of-state models to learn how well the pipeline can
differentiate the various types of gravitational waves.
Data Quality for Binary Black Hole Searches With aLIGO
Albert Kamau Wandui
Mentors: Alan Weinstein and Jonah Kanner
The aLIGO detectors are expected to reach an unprecedented level of sensitivity, which will make direct
observations of gravitational waves possible for the very first time. The search for gravitational wave signals, in
particular, from the inspiral, merger and ringdown of black hole binaries has been limited by the presence of noise
transients called glitches, which trigger in the gravitational strain channel. We describe a technique for vetoing
these glitches by finding correlations between possible glitches in our Binary Black Hole (BBH) Pipeline and noise
transients identified in auxiliary channels of the aLIGO instrument. This involves finding the coincidence window
between the triggers in the BBH pipeline and triggers in the auxiliary channels that maximizes the efficiency of
eliminating glitches from the BBH pipeline and at the same time has a minimal impact on the detector live time.
The goal of this technique is to remove the non-gaussian instrumental transients seen in the gravitational wave
strain data, which will enable better determination of detection thresholds for actual gravitational wave signals
from binary black holes.

108

Numerical Simulations of Black-Hole Binaries

Lindsey Whitesides
Mentor: Mark Scheel
The inspiral and merger of binary black-hole systems produce gravitational waves which Advanced LIGO aims to
detect. Numerical simulations of these binaries are used to compute waveform templates that help LIGO pull
signals out of noise and determine binary parameters such as masses and spins. We are conducting simulations of
binary black-hole mergers with high spins of 0.91 and 0.99 oriented in a superkick configuration, in which linear
momentum is radiated anisotropically and the remnant black-hole acquires a velocity of up to thousands of km/s.
The final velocity depends sensitively on the initial spin orientations. Once our simulations complete, we will
determine whether LIGO can distinguish between initial spin orientations for these high-spin systems; it cannot for
low-spin systems. We have also developed new code which measures the eccentricity of the orbit and computes
new orbital parameters. It is written in C++, and includes error bars on computed quantities. Several simulations
are run iteratively, adjusting the orbital parameters each time until the target eccentricity is reached. Our new code
will run during the main simulation instead of afterwards, allowing each iteration to end earlier, and thus increasing
the speed of these simulations.
Investigation of Thermal Noise in Thin Silicon Structures
Matthew Winchester
Mentors: Nicolas Smith, Zach Korth, and Rana Adhikari
Current aLIGO (Advanced Laser Interferometer Gravitational-Wave Observatory) suspensions and test masses are
built from a fused silica substrate. In an effort to further increase detector sensitivity in the mid LIGO frequency
band, which is currently limited by thermal noise, cryogenic silicon has become a candidate for the next generation
of detector suspensions and test masses due to its excellent mechanical and optical properties. The fluctuationdissipation theorem links microscopic thermal noise fluctuations with macroscopic material damping, which in turn
motivates the study of damping mechanisms in silicon structures. In this project we demonstrate and assess
several methods for measuring the quality factor of silicon cantilevers, including a continuous measurement
technique capable of measuring the quality factor of several resonant modes simultaneously. We also investigate
the effects of parameters such as temperature, cantilever geometry, and surface treatments on the quality factor
with the goal of informing future detector suspension designs.
Detecting Gravitational Wave Transients From Unmodeled Sources
Enia Xhakaj
Mentors: Jonah Kanner and Alan Weinstein
With the recent installation of the Advanced LIGO detectors in Hanford and Livingston, the sensitivity of detecting
gravitational waves (GWs) has improved significantly. However, especially for unmodeled sources, it is still difficult
to distinguish signals from instrumental noise with high confidence. The recent science and engineering runs
suggest that apart from stationary Gaussian noise there are other instrument artifacts, called "glitches" that
impose a challenge in recovering signals. The BayesWave Pipeline presents a novel way to approach this problem.
BayesWave uses modern statistical methods to model the data as signals or glitches and picks the most likely
model as the representative for the data. In this project, we test the BayesWave performance of recovering
injected theoretical signals in the LIGO Sixth Science run (S6). We run BayesWave as a follow-up to coherent
WaveBurst (cWB), which is the primary analysis pipeline for identifying burst signals in noisy data from the LIGO
detectors, and we compare the efficiency for confident burst detection using cWB alone with that of the combined
pipeline.
Carrier Independent IFO Cavity Control Scheme
Weiming Zhao
Mentor: Dick Gustafson
We analyze, build, demonstrate, and characterize a sideband-on-sideband variant of the Pound-Drever-Hall scheme
for controlling an interferometer that is independent of the carrier frequency. This should offer more robust control
of the lock acquisition of multiple Fabry-Perot cavities. The novel technique requires phase modulated sidebands to
be also resonant in the cavity. Non-resonant amplitude modulated sidebands will require an acousto-optic
modulator that creates a second set of sidebands that are used to lock the phase modulated sidebands. This
promises graceful control of Michelson cavities during carrier locking of the arm cavities of an interferometer.

109

Abstracts
StudentFaculty Programs

2
0
1
5