∗

A Quantum Engineer’s Guide to Superconducting Qubits

P. Krantz1,2,† , M. Kjaergaard1 , F. Yan1 , T.P. Orlando1 , S. Gustavsson1 , and W. D. Oliver1,3,‡
1
Research Laboratory of Electronics, Massachusetts Institute of Technology, Cambridge, MA 02139,
USA
2
Wallenberg Centre for Quantum Technology (WACQT), Chalmers University of Technology, Gothenburg, SE-41296,
Sweden and
3
MIT Lincoln Laboratory, 244 Wood Street, Lexington, MA 02420, USA

(Dated: 9 July 2021)

The aim of this review is to provide quantum engineers with an introductory guide to the central concepts
and challenges in the rapidly accelerating field of superconducting quantum circuits. Over the past twenty
years, the field has matured from a predominantly basic research endeavor to one that increasingly explores
the engineering of larger-scale superconducting quantum systems. Here, we review several foundational
arXiv:1904.06560v5 [quant-ph] 7 Jul 2021

elements – qubit design, noise properties, qubit control, and readout techniques – developed during this
period, bridging fundamental concepts in circuit quantum electrodynamics (cQED) and contemporary, state-
of-the-art applications in gate-model quantum computation.

CONTENTS 1. Connecting T1 to S(ω) 19

2. Connecting Tϕ to S(ω) 20
I. Introduction 2 3. Noise spectroscopy 21
A. Organization of this article 2 E. Engineering noise mitigation 21
1. Materials and fabrication improvements 22
II. Engineering quantum circuits 3 2. Design improvements 22
A. From quantum harmonic oscillator to the 3. Dynamical error suppression 22
transmon qubit 3 4. Cryogenic engineering 22
B. Qubit Hamiltonian engineering 6
1. Tunable qubit: split transmon 6 IV. Qubit control 22
2. Towards larger anharmonicity: flux qubit A. Boolean logic gates used in classical
and fluxonium 7 computers 23
C. Interaction Hamiltonian engineering 9 B. Quantum logic gates used in quantum
1. Physical coupling: capacitive and computers 24
inductive 9 C. Comparing classical and quantum gates 26
2. Coupling axis: transverse and 1. Gate sets and gate synthesis 27
longitudinal 10 2. Addressing superconducting qubits 27
D. Single-qubit gates 27
III. Noise, decoherence, and error mitigation 12 1. Capacitive coupling for X,Y control 28
A. Types of noise 12 2. Virtual Z gate 30
1. Systematic noise 12 3. The DRAG scheme 30
2. Stochastic noise 12 E. The iSWAP two-qubit gate in tunable
3. Noise strength and qubit susceptibility 12 qubits 31
B. Modeling noise and decoherence 12 1. Deriving the iSWAP unitary 31
1. Bloch sphere representation 12 2. Applications of the iSWAP gate 33
2. Bloch-Redfield model of decoherence 13 F. The CPHASE two-qubit gate in tunable
3. Modification due to 1/f -type noise 16 qubits 33
4. Noise power spectral density (PSD) 17 1. Trajectory design for the CPHASE gate 35
C. Common examples of noise 18 2. The CPHASE gate for quantum error
1. Charge noise 18 correction 35
2. Magnetic flux noise 18 3. Quantum simulation and algorithm
3. Photon number fluctuations 19 demonstrations using CPHASE 36
4. Quasiparticles 19 G. Two-qubit gates using only microwaves 37
D. Operator form of qubit-environment 1. The operational principle of the CR gate 37
interaction 19 2. Improvements to the CR gate and
quantum error correction experiments
using CR 38
3. Quantum simulation and algorithm
∗† [email protected],‡ [email protected]
demonstrations with the CR gate 38
 2

4. Other microwave-only gates: bSWAP, and polarized photons30–33 , where the quantum informa-
MAP, and RIP 39 tion is encoded in natural microscopic quantum systems,
H. Gate implementations with tunable superconducting qubits are macroscopic in size and litho-
coupling 40 graphically defined.
One remarkable feature of superconducting qubits is
V. Qubit readout 41 that their energy-level spectra are governed by circuit el-
A. Dispersive readout 41 ement parameters and thus are configurable; they can be
B. Measuring the resonator amplitude and designed to exhibit “atom-like” energy spectra with de-
phase 43 sired properties. Therefore, superconducting qubits are
1. Representation of the readout signal 43 also often referred to as artificial atoms, offering a rich
2. I-Q mixing 44 parameter space of possible qubit properties and opera-
3. Homodyne demodulation 44 tion regimes, with predictable performance in terms of
transition frequencies, anharmonicity, and complexity.
4. Heterodyne demodulation 45
While there are many other excellent reviews on su-
C. Weak and strong qubit measurements:
perconducting qubits, see e.g. Refs. 34–43, this work
Impact of noise 46
specifically aims to introduce new quantum engineers
D. “Purcell filters” for faster readout 48
(academic and industrial alike) to the terminology and
E. Improve signal-to-noise ratio: Parametric state-of-the-art practices used in the rapidly accelerat-
amplification 50 ing field of superconducting quantum computing. The
1. Quantum-limited amplification processes 50 reader is assumed to be familiar with basic concepts that
2. Operation of Josephson parametric span classical physics, quantum mechanics, and electrical
amplifiers 51 engineering. In particular, readers will find it useful to
3. The traveling wave parametric amplifier 53 have had previous exposure to classical mechanics, the
Schrödinger equation, the Bloch sphere representation of
VI. Summary and outlook 54 qubit states, second quantization, basic concepts of su-
perconductivity, electromagnetism, introductory circuit
Acknowledgments 55 analysis, classical boolean logic, linear dynamical sys-
tems, analog and digital signal processing, and famil-
iarity with microwave components such as transmission
I. INTRODUCTION lines and mixers. These topics will be introduced as they
arise, but having basic prior knowledge will be helpful.
Quantum processors harness the intrinsic properties of
quantum mechanical systems – such as quantum paral-
lelism and quantum interference – to solve certain prob- A. Organization of this article
lems where classical computers fall short1–6 . Over the
past two decades, rapid developments in the science and This review is organized in the following four sections;
engineering of quantum systems have advanced the fron- first, in Sec. II, we explore the parameter space avail-
tier in quantum computation, from the realm of scien- able when designing superconducting circuits. In particu-
tific explorations on single isolated quantum systems to- lar, we look at the promising capacitively-shunted planar
ward the creation and manipulation of multi-qubit pro- qubit modalities and how these can be engineered with
cessors7,8 . In particular, the requirements imposed by desired properties, such as transition frequency, anhar-
larger quantum processors have shifted of mindset within monicity, and reduced susceptibility to various sources
the community, from solely scientific discovery to the de- of noise. In this section, we also introduce several ways
velopment of new, foundational engineering abstractions in which interactions between qubits can be engineered,
associated with the design, control, and readout of multi- in order to implement two-qubit entangling operations,
qubit quantum systems. The result is the emergence of a needed for a universal gate set.
new discipline termed quantum engineering, which serves In Sec. III, we discuss systematic and stochastic noise,
to bridge the basic sciences, mathematics, and computer the concepts of noise strength and qubit noise suscepti-
science with fields generally associated with traditional bility, and the common sources of noise which lead to
engineering. decoherence in superconducting circuits. We introduce
One prominent platform for constructing a multi-qubit the Bloch-Redfield model of decoherence, characterized
quantum processor involves superconducting qubits, in by longitudinal and transverse relaxation times T1 and
which information is stored in the quantum degrees of T2 , and discuss the implications of 1/f noise. We then
freedom of nanofabricated, anharmonic oscillators con- define the noise power spectral density, which is com-
structed from superconducting circuit elements. In con- monly used to characterize noise processes, and describe
trast to other platforms, e.g. electron spins in sili- how it drives decoherence. Finally, we close the section
con9–14 and quantum dots15–18 , trapped ions19–23 , ul- with a review of coherent control methods used to miti-
tracold atoms24–27 , nitrogen-vacancies in diamonds28,29 , gate certain types of coherence, reversible noise.
 3

In Sec. IV, we provide a review of how single- and The time-independent Hamiltonian Ĥ governs the time
two-qubit operations are typically implemented in super- evolution of the system through the operator e−iĤt/~ .
conducing circuits, by using a combination of local mag- Thus, just as with classical systems, determining the
netic flux control and microwave drives. In particular, we Hamiltonian of a system – whether the classical Hamil-
discuss the family of two-qubit gates arising from a ca- tonian H or its quantum counterpart Ĥ – is the first step
pacitive coupling between qubits, and introduce several to deriving its dynamical behavior. In Sec. IV, we con-
recent advances that have been demonstrated to achieve sider the case when the Hamiltonian is time-dependent
high-fidelity gates, as well as applications in quantum in the context of qubit control.
information processing that use these gates. The contin- To understand the dynamics of a superconducting
ued development of high-fidelity two-qubit gates in super- qubit circuit, it is natural to start with the classical de-
conducting qubits is a highly active research area. For scription of a linear LC resonant circuit [Fig. 1(a)]. In
this reason, we include sufficient technical details that a this system, energy oscillates between electrical energy
reader may use this review as a starting point to critically in the capacitor C and magnetic energy in the inductor
assess the pros and cons of the various gates, as well as L. In the following, we will arbitrarily associate the elec-
develop an appreciation for the types of gate-engineering trical energy with the “kinetic energy” and the magnetic
already implemented in state-of-the-art superconducting energy with the “potential energy” of the oscillator. The
quantum processors. instantaneous, time-dependent energy in each element is
Finally, in Sec. V, we discuss the physics and engi- derived from its current and voltage,
neering associated with the dispersive readout technique, Z t
typically used to measure the individual qubit states in E(t) = V (t0 )I(t0 )dt0 , (3)
modern quantum processors. After a discussion of the −∞
theory behind dispersive coupling, we give an introduc-
tion to design of Purcell filters and the development of where V (t ) and I(t ) denote the voltage and current of
0 0

quantum-limited parametric amplifiers. the capacitor or inductor.

To derive the classical Hamiltonian, we follow the stan-
dard approach used in classical mechanics: the Lagrange-
II. ENGINEERING QUANTUM CIRCUITS Hamilton formulation. Here, we represent the circuit el-
ements in terms of one of its generalized circuit coor-
In this section, we will demonstrate how quantum sys- dinates, charge or flux. In the following, we pick flux,
tems based on superconducting circuits can be engineered defined as the time integral of the voltage
to achieve certain desired properties. Using the most
common qubit modalities, we discuss how properties such t
Z
as the qubit transition frequency, anharmonicity, and Φ(t) = V (t0 )dt0 . (4)
−∞
noise susceptibility can be tailored by the choice of circuit
topology and element parameter values. We also discuss In this example, the voltage at the node is also the branch
how to engineer the interactions between different quan- voltage across the element. In this section, we will simply
tum systems, in particular the cases of qubit-qubit and refer to these as node voltages and fluxes for convenience.
qubit-resonator couplings. For a more detailed discussion of nodes and branches in
this context, we refer the reader to Ref. 44.
Note that in the following, we could have exchanged
A. From quantum harmonic oscillator to the transmon our associations with kinetic energy (momentum coor-
qubit dinate) and potential energy (position coordinate), and
instead start with the charge variable Q(t), which is the
A quantum mechanical system is governed by the time- time integral of the current I(t).
dependent Schrödinger equation, By combining Eqs. (3) and (4), using the relations V =
L dI/dt and I = C dV /dt, and applying the integration
∂ by parts formula, we can write down energy terms for the
Ĥ|ψ(t)i = i~ |ψ(t)i, (1)
∂t capacitor and inductor in terms of the node flux,
where |ψ(t)i is the state of the quantum system at time
t, ~ is the reduced Planck’s constant h/2π, and Ĥ is the 1
TC = C Φ̇2 , (5)
Hamiltonian that describes the total energy of the sys- 2
tem. The “hat” is used to indicate that Ĥ is a quantum
operator. As the Schrödinger equation is a first-order lin- 1 2
ear differential equation, the temporal dynamics of the UL = Φ . (6)
quantum system may be viewed as a straightforward ex- 2L
ample of a linear dynamical system with formal solution, The Lagrangian is defined as the difference between
the kinetic and potential energy terms and can thus be
|ψ(t)i = e−iĤt/~ |ψ(0)i. (2) expressed in terms of Eqs. (5) and (6)
 4

across the inductor. These two operators form a canon-

ical conjugate pair, obeying the commutation relation
1 1 2
L = TC − U L = C Φ̇2 − Φ . (7) [φ, n] = i. We note that the factor 4 in front of the
2 2L charging energy EC is solely a historical artifact, namely,
From the Lagrangian in Eq. (7), we can further derive that this energy scale was first defined for single-electron
the Hamiltonian using the Legendre transformation, for systems and then adopted to two-electron Cooper-pair
which we need to calculate the momentum conjugate to systems.
the flux, which in this case is the charge on the capacitor The Hamiltonian in Eq. (13) is identical to the one de-
scribing a particle in a one-dimensional quadratic poten-
tial, a quantum harmonic oscillator (QHO). We can treat
∂L φ as the generalized position coordinate, so that the first
Q= = C Φ̇. (8)
∂ Φ̇ term is the kinetic energy and the second term is the po-
tential energy. We emphasize that the functional form
The Hamiltonian of the system is now defined as
of the potential energy influences the eigensolutions. For
example, the fact that this term is quadratic (UL ∝ φ2 )
Q2 Φ2 1 1 in Eq. (13) gives rise to the shape of the potential in Fig.
H = QΦ̇ − L = + ≡ CV 2 + LI 2 , (9) 1(b). The solution to this eigenvalue problem gives an in-
2C 2L 2 2
finite series of eigenstates |ki, (k = 0, 1, 2, . . .), whose cor-
as one would expect for an electrical LC circuit. Note responding eigenenergies Ek are all√equidistantly spaced, √
that this Hamiltonian is analogous to that of a mechan- i.e. Ek+1 −Ek = ~ωr , where ωr = 8EL EC /~ = 1/ LC
ical harmonic oscillator,√with mass m = C, and reso- denotes the resonant frequency of the system, see Fig.
nant frequency ω = 1/ LC, which expressed in posi- 1(b). We may represent these results in a more compact
tion, x, and momentum, p, coordinates takes the form form (second quantization) for the quantum harmonic
H = p2 /2m + mω 2 x2 /2. oscillator (QHO) Hamiltonian
The Hamiltonian described above is classical. In order
to proceed to a quantum-mechanical description of the
1
 
system, we need to promote the charge and flux coordi- H = ~ωr a a +
†
, (14)
nates to quantum operators. And, whereas the classical 2
coordinates satisfy the Poisson bracket:
where a† (a) is the creation (annihilation) operator of a
δf δg δg δf single excitation of the resonator. The Hamiltonian in
{f, g} = − (10) Eq. (14) is written as an energy. It is, however, often
δΦ δQ δΦ δQ
δΦ δQ δQ δΦ preferred to divide by ~ so that the expression has units
→ {Φ, Q} = − = 1 − 0 = 1, (11) of radian frequency, since we will later resonantly drive
δΦ δQ δΦ δQ
transitions at a particular frequency or reference the rate
the quantum operators similarly satisfy a commutation at which two systems interact with one another. There-
relation: fore, from here on, ~ will be omitted.
The original charge number and phase operators can
[Φ̂, Q̂] = Φ̂Q̂ − Q̂Φ̂ = i~, (12) be expressed as n = nzpf ×i(a−a† ) and φ = φzpf ×(a+a† ),
where nzpf = [EL /(32EC )]1/4 and φzpf = (2EC /EL )1/4
where the operators are indicated by hats. From this are the zero-point fluctuations of the charge and phase
point forward, however, the hats on operators will be variables, respectively. Quantum mechanically, the quan-
omitted for simplicity. tum states are represented as wavefunctions that are gen-
In a simple LC resonant circuit [Fig. 1(a)], both the erally distributed over a range of values of n and φ and,
inductor L and the capacitor C are linear circuit ele- consequently, the wavefunctions have non-zero standard
ments. Defining the reduced flux φ ≡ 2πΦ/Φ0 and the deviations. Such wavefunction distributions are referred
reduced charge n = Q/2e, we can write down the follow- to as “quantum fluctuations,” and they exist, even in the
ing quantum-mechanical Hamiltonian for the circuit, ground state, where they are called “zero-point fluctua-
tions”.
1 The linear characteristics of the QHO has a natural
H = 4EC n2 + EL φ2 , (13) limitation in its applications for processing quantum in-
2
formation. Before the system can be used as a qubit, we
where EC = e2 /(2C) is the charging energy required need to be able to define a computational subspace con-
to add each electron of the Cooper-pair to the island sisting of only two energy states (usually the two-lowest
and EL = (Φ0 /2π)2 /L is the inductive energy, where energy eigenstates) in between which transitions can be
Φ0 = h/(2e) is the superconducting magnetic flux quan- driven without also exciting other levels in the system.
tum. Moreover, the quantum operator n is the ex- Since many gate operations, such as single-qubit gates
cess number of Cooper-pairs on the island, and φ – the (Sec. IV), depend on frequency selectivity, the equidis-
reduced flux – is denoted the “gauge-invariant phase” tant level-spacing of the QHO, illustrated in Fig. 1(b),
 5

(a) i (c)
+ LJ ~ dφ
Lr Cr v Cs I = Ic sin(φ), V = , (15)
CJ 2e dt
-
resulting in a modified Hamiltonian
(b) (d)
5 5
Transmon
4 4
H = 4EC n2 − EJ cos(φ), (16)

]
]

3 3
Energy [

Energy [
where EC = e2 /(2CΣ ), CΣ = Cs + CJ is the total ca-
2 2
pacitance, including both shunt capacitance Cs and the

subspace
Comp.
1 1 self-capacitance of the junction CJ , and EJ = Ic Φ0 /2π is
0
QHO the Josephson energy, with Ic being the critical current
0
- - /2 0 /2 - - /2 0 /2 of the junction‡ . After introducing the Josephson junc-
Superconducting phase, Superconducting phase, tion in the circuit, the potential energy no longer takes
a manifestly parabolic form (from which the harmonic
FIG. 1. (a) Circuit for a parallel LC-oscillator (quantum har- spectrum originates), but rather features a cosinusoidal
monic oscillator, QHO), with inductance L in parallel with form, see the second term in Eq. (16), which makes the
capacitance, C. The superconducting phase on the island is energy spectrum non-degenerate. Therefore, the Joseph-
denoted φ, referencing ground as zero. (b) Energy potential son junction is the key ingredient that makes the oscilla-
for the QHO, where energy levels are equidistantly spaced tor anharmonic and thus allows us to identify a uniquely
~ωr apart. (c) Josephson qubit circuit, where the nonlinear addressable quantum two-level system, see Fig. 1(d).
inductance LJ (represented with the Josephson-subcircuit in Once the nonlinearity has been added, the system dy-
the dashed orange box) is shunted by a capacitance, Cs . (d) namics is governed by the dominant energy in Eq. (16),
The Josephson inductance reshapes the quadratic energy po-
reflected in the EJ /EC ratio. Over time, the super-
tential (dashed red) into sinusoidal (solid blue), which yields
non-equidistant energy levels. This allows us to isolate the
conducting qubit community has converged towards cir-
two lowest energy levels |0i and |1i, forming a computational cuit designs with EJ  EC . In the opposite case when
subspace with an energy separation ~ω01 , which is different EJ ≤ EC , the qubit becomes highly sensitive to charge
than ~ω12 . noise, which has proven more challenging to mitigate
than flux noise, making it very hard to achieve high co-
herence. Another motivation is that current technologies
allow for more flexibility in engineering the inductive (or
poses a practical limitation† . potential) part of the Hamiltonian. Therefore, working
To mitigate the problem of unwanted dynamics in- in the EJ ≤ EC limit, makes the system more sensitive
volving non-computational states, we need to add anhar- to the change in the potential Hamiltonian. Therefore,
monicity (or nonlinearity) into our system. In short, we we will focus here on the state-of-the-art qubit modalities
require the transition frequencies ωq0→1 and ωq1→2 be suffi- that fall in the regime EJ  EC . For readers who are
ciently different to be individually adressable. In general, interested in the physics in the EJ ≤ EC regime, such
the larger the anharmonicity the better. In practise, the as the earlier Cooper-pair box charge qubit, we refer to
amount of anharmonicity sets a limit on how short the Refs. 48–51.
pulses used to drive the qubit can be. This is discussed To access the EJ  EC regime, one preferred approach
in detail in Sec. IV D 3. is to make the charging EC small by shunting the junction
To introduce the nonlinearity required to modify the with a large capacitor, Cs  CJ , effectively making the
harmonic potential, we use the Josephson junction – a qubit less sensitive to charge noise – a circuit commonly
nonlinear, dissipationless circuit element that forms the known as the transmon qubit52 . In this limit, the super-
backbone in superconducting circuits46,47 . By replacing conducting phase φ is a good quantum number, i.e. the
the linear inductor of the QHO with a Josephson junc- spread (or quantum fluctuation) of φ values represented
tion, playing the role of a nonlinear inductor, we can by the quantum wavefunction is small. The low-energy
modify the functional form of the potential energy. The eigenstates are therefore, to a good approximation, local-
potential energy of the Josephson junction can be derived ized states in the potential well, see Fig. 1(d). We may
from Eq. (3) and the two Josephson relations gain more insight by expanding the potential term of Eq.
(16) into a power series (since φ is small), that is

† Even though linear resonant systems cannot be addressed properly,

their long coherence times have proven them useful as quantum ac-
cess memories for storing quantum information, where a nonlinear ‡ The critical current is the maximum supercurrent that the junction
ancilla system is used as a quantum controller for feeding and ex- can support before it switches to the resistive state with non-zero
tracting excitations to/from the resonant cavity modes45 . voltage.
 6

B. Qubit Hamiltonian engineering

1 1
EJ cos(φ) = EJ φ2 − EJ φ4 + O(φ6 ). (17) 1. Tunable qubit: split transmon
2 24
The leading quadratic term in Eq. (17) alone will To implement fast gate operations with high-fidelity,
result in a QHO, recall Eq. (13). The second term, as needed to implement quantum logic, many (though
however, is quartic which modifies the eigensolution and not all63 ) of the quantum processor architectures imple-
disrupts the otherwise harmonic energy structure. Note mented today feature tunable qubit frequencies64–67 . For
that, the negative coefficient of the quartic term indicates instance, in some cases, we need to bring two qubits into
that the anharmonicity α = ωq1→2 − ωq0→1 is negative and resonance to exchange (swap) energy, while we also need
its limit in magnitude thus cannot be made arbitrarily the capability of separating them during idling periods
large. For the case of the transmon, α = −EC is usually to minimize their interactions. To do this, we need an
designed to be 100 − 300 MHz, as √ required to maintain external parameter which allows us to access one of the
a desirable qubit frequency ωq = ( 8EJ EC − EC )/~ = degrees of freedom of the system in a controllable fashion.
3−6 GHz, while keeping an energy ratio sufficiently large One widely-used technique is to replace the single
(EJ /EC ≥ 50) to suppress charge sensitivity52 . Fortu- Josephson junction with a loop interupted by two iden-
nately, the charge sensitivity is exponentially suppressed tical junctions – forming a dc superconducting quantum
for increased EJ /EC , while the reduction in anharmonic- interference device (dc-SQUID)68 . Due to the interfer-
ity only scales as a weak power law, leading to a workable ence between the two arms of the SQUID, the effective
device. critical current of the two parallel junctions can be de-
Including terms up to fourth order and using the QHO creased by applying a magnetic flux threading the loop,
eigenbases, the system Hamiltonian resembles that of a see Fig. 2(a). Due to the fluxoid quantization condition,
Duffing oscillator the algebraic sum of branch flux of all of the inductive
elements along the loop plus the externally applied flux
α † † equal an integer number of superconducting flux quanta,
H = ωq a† a + a a aa. (18) that is
2
Since |α|  ωq , we can see that the transmon qubit is
basically a weakly anharmonic oscillator (AHO). If exci- ϕ1 − ϕ2 + 2ϕe = 2πk, (20)
tation to higher non-computational states is suppressed
over any gate operations, either due to a large enough |α| where ϕe = πΦext /Φ0 . Using this condition, we can elim-
or due to robust control techniques such as the DRAG inate one degree of freedom and treat the SQUID-loop
pulse, see Sec. IV D 3, we may effectively treat the AHO as a single junction, but with the important modification
as a quantum two-level system, simplifying the Hamilto- that EJ is tunable (via the SQUID critical current) by
nian to means of the external flux Φext . The effective Hamilto-
nian of the so-called split transmon (ignoring the con-
σz stant) is
H = ωq , (19)
2
where σz is the Pauli-z operator. However, one should H = 4EC n2 − 2EJ |cos (ϕe )| cos(φ). (21)
always keep in mind that the higher levels physically | {z }
exist53 . Their influence on system dynamics should be EJ0 (ϕe )

taken into account when designing the system and its

control processes. In fact, there are many cases where We can see that Eq. (21) is analogous to Eq. (16), with
the higher levels have proven useful to implement more EJ replaced by EJ0 (ϕe ) = 2EJ |cos (ϕe )|. The magnitude
efficient gate operations54 . of the net, effective Josephson energy EJ0 has a period
In addition to reducing the charge dispersion, the use of Φ0 in applied flux and spans from 0 to its maximum
of a large shunt capacitor also enables us to engineer value 2EJ . Therefore, the qubit frequency can be tuned
the electric field distribution of the quantum system, and periodically with Φext , see Fig. 2(b).
thus the participation of surface loss mechanisms. In the While the split transmon enables frequency tunabil-
development of the 3D transmon55 , e.g. a 2D transmon ity by the externally applied magnetic field, it also in-
coupled to a 3D cavity, it was demonstrated that by mak- troduces sensitivity to random flux fluctuations, known
ing the gap between the two lateral capacitor plates large as flux noise. At any working point, the slope of the
(compared to the film thickness) the coherence time in- qubit spectrum, ∂ωq /∂Φext , indicates to first order how
creases since a smaller portion of the electric field in- strongly this flux noise affects the qubit frequency. The
teracts with the lossy interfaces, e.g. metal-substrate sensitivity is generally non-zero, except at multiples of
and substrate-vacuum interfaces, which has been stud- the flux quantum, Φext = kΦ0 , where k is an integer,
ied extensively56–61 . where ∂ωq /∂Φext = 0.
 7

(a) Symmetric transmon (c) Asymmetric transmon (e) C-shunted Flux qubit (g) C-shunted Fluxonium

φ1 φ1
φ2
φ1 φ2 φ1 φ2 CJ
Ic Ic Ic φ2 φN

(b) (d) (f) (h)

8 8 8 8
Ec /h = 0.3 GHz
EJ /h = 15 GHz
6 6 6 6

4 4 4 4

2 2 2 2

0 0 0 0
- - /2 0 /2 - - /2 0 /2 - - /2 0 /2 - - /2 0 /2

FIG. 2. Modular qubit circuit representations for capacitively shunted qubit modalities (orange box Fig. 1c) and corresponding
qubit transition frequencies for the two lowest energy states as a function of applied magnetic flux in units of Φ0 . (a-b)
Symmetric transmon qubit, with Josephson energy EJ are shunted with a capacitor yielding a charging energy EC . (c-d)
Asymmetric transmon qubit, with junction asymmetry γ = EJ2 /EJ1 = 2.5. (e-f) Capacitively shunted flux qubit, where a
small principle junction (red) is shunted with two larger junctions (orange). Parameters are the same as Yan et al.62 . (g-h)
C-shunted fluxonium qubit, where the small junction is inductively shunted with a large array of N junctions.

One recent development has focused on reducing the resonance gate is optimized with certain frequency de-
qubit sensitivity to flux noise, while maintaining suf- tuning between two qubits70 . Therefore, by using an
ficient tunability to operate our quantum gates. The asymmetric transmon, a small frequency-tuning range
idea is to make the two junctions in the split transmon is introduced that is sufficient to compensate for fabri-
asymmetric69 , see Fig. 2(c). This yields the following cation variations, without introducing unnecessary large
Hamiltonian susceptibility to flux noise and thus maintaining high co-
herence. For another example, a surface code scheme
q based on the adiabatic CPHASE-gate requires specific fre-
H = 4EC n2 − EJΣ cos2 (ϕe ) + d2 sin2 (ϕe ) cos(φ), (22) quency configuration among qubits in order to avoid fre-
| {z } quency crowding issues, and asymmetric transmons fit
EJ0 (ϕe )
well with its well-defined frequency range71 . In general,
where EJΣ = EJ1 + EJ2 and d = (γ − 1)/(γ + 1) is as the quantum processors scale up and fabrication im-
the junction asymmetry parameter, with γ = EJ2 /EJ1 . proves, asymmetric transmons are likely to be found in
Again, we can treat the two junctions as a single-junction wider applications in the future.
transmon, with an effective Josephson energy EJ0 (ϕe ). In
particular, we can recognize the two special cases; for
d = 0, the Hamiltonian in Eq. (22) reduces to the sym- 2. Towards larger anharmonicity: flux qubit and fluxonium
metric case with EJ0 (ϕe ) = EJΣ |cos(ϕe )|, as in Eq. (21)
with EJΣ = 2EJ . In the other limit, when |d| → 1, We see that split transmon qubits, be it symmetric or
EJ0 (ϕe ) → EJΣ and the flux-tunability of the Josephson not, still share the same topology as the single junction
energy vanishes, which is equivalent to the single junction version, yielding a sinusoidal potential. Therefore, the
case, recall Eq. (16). degree to which the properties of these qubits can be en-
From the discussion above we see that going from sym- gineered has not fundamentally changed. In particular,
metric to asymmetric transmons does not change the cir- the limited anharmonicity in transmon-type qubits in-
cuit topology. This seemingly trivial modification, how- trinsically causes significant residual excitation to higher-
ever, has profound impact for practical applications. As energy states, undermining performance of gate opera-
we can see from the qubit spectra, Fig. 2(d), the flux tions. To go beyond this, it is necessary to introduce
sensitivity is suppressed across the entire tunable fre- additional complexity into the circuit.
quency range. For example, the performance of the cross- One outstanding development in this regard is the in-
 8

vention of the flux qubit72,73 , where the qubit loop is in- the introduction of the capacitive shunt, similar to the
terrupted by three (or four) junctions, see Fig. 2(e). On modified Cooper-pair box leading to the transmon qubit.
one branch is one smaller junction; on the other branch The double-well case obtained for γ < 2 was demon-
are two identical junctions, both a factor γ larger in size strated and investigated much earlier72,73 . The intuitive
compared to the small junction. The addition of one picture based on circulating current states – so it gets
more junction as compared to the split transmon is non- the name persisting-current flux qubit (PCFQ) – gives
trivial, as it changes the circuit topology and reshapes a satisfying physical description of the qubit degrees of
the potential energy profile. freedom. However, from the perspective of a quantum
Each junction is associated with a phase variable, and engineer, the qubit properties are of more interest, even
the fluxoid quantization condition again allows us to if sometimes we may lose physical intuition about the
eliminate one degree of freedom. Consequently, we have system in certain regimes; such as when γ ≈ 2 and there
a two-dimensional potential landscape, which in compar- are no clear circulating current states. The most impor-
ison to the simpler topology of the transmon, compli- tant feature of the PCFQ is that its anharmonicity can
cates the problem both conceptually and computation- be much greater than the transmon and CSFQ and the
ally. Fortunately, under the assumed setting that the ar- transition matrix elements |h1|n̂|0i|, |h1|φ̂|0i| become con-
ray junctions are larger in size (γ > 1), it is usually a good siderably smaller given equivalent EJ /EC . Therefore, a
approximation to treat the problem as a particle moving longer relaxation time can be expected. These features
in a quasi-1D potential, which also helps us gain more have been demonstrated even more prominently in its
insight and intuition about the system and draw qualita- close relative, the fluxonium qubit75 .
tive conclusions. The Hamiltonian under this quasi-1D The flux qubit is a striking example that illustrates
approximation reads, how one dramatically can engineer the qubit properties
through the choice of various circuit parameters. The in-
troduction of array junctions and consequent biharmonic
H ≈ 4EC n2 − EJ cos(2φ + ϕe ) − 2γEJ cos(φ). (23) profile generates rich dynamics as well as broad applica-
tions. An extention of this idea is the fluxonium qubit,
Note that the phase variable in Eq. (23) is the sum which generated substantial interest recently, due partly
of the branch phases across the two array junctions, φ = to its capability of engineering the transition matrix el-
(ϕ1 + ϕ2 )/2, assuming the same current direction across ements to achieve millisecond T1 time, and due partly
ϕ1 and ϕ2 . The external magnetic flux is denoted ϕe = to the invention of novel gate schemes applicable to such
2πΦext /Φ0 . The second term in Eq. (23) is contributed well-protected qubits76,77 .
by the small junction with Josephson energy EJ , whereas Compared to flux qubits, which usually contain two
the third term takes into account the two array junctions, or three array junctions78 , the number of array junctions
together with Josephson energy 2γEJ . Clearly, the sum in the fluxonium qubit is dramatically increased75,79 , in
of these two terms no longer has the characteristics of a some cases, to the order of 100, see Fig. 2(g). Following
simple cosinusoid, and the final potential profile as well the same quasi-1D approximation as for the flux qubit,
as the corresponding eigenstates depends on both the the last term in Eq. (23) becomes −N γEJ cos(φ/N ),
external flux ϕe and the junction area ratio γ. where N denotes the number of array junctions. For
large N , the argument in the cosine term φ/N becomes
The most common working point for this system is
sufficiently small that a second order expansion is a good
when ϕe = π + 2πk, where k is an integer – that is
approximation. This results in the fluxonium Hamilto-
when half a superconducting flux quantum threads the
nian,
qubit loop. At this flux bias point, the qubit spectrum
reaches its minimum, and the qubit frequency is first-
order insensitive to flux noise, see Fig. 2(f). This point 1
H ≈ 4EC n2 − EJ cos(φ + ϕe ) + EL φ2 , (24)
is often referred to as the flux degeneracy point, where 2
flux qubits tend to have the optimal coherence time. where EL = (γ/N )EJ is the inductive energy of the
At this operation point, the potential energy may as- effective inductance contributed by the junction ar-
sume a single-well (γ ≥ 2) or a double-well (γ < 2) ray – often known as superinductance due to its large
profile. The single-well case shares some simularities value79–81 . Therefore, we can treat the potential energy
with the transmon qubit, where the quadratic and quar- as a quadratic term modulated by a sinusoidal term, sim-
tic terms of the Hamiltonian determines the harmonic- ilar to that of an rf-SQUID type flux qubit82 . However,
ity and anharmonicity, respectively. The capacitively- the kinetic inductance of the Josephson junction array is
shunted flux qubit (CSFQ)62,74 was explored in this in general much larger than the geometric inductance of
regime, demonstrating long coherence and decently high the wire in an rf-SQUID.
anharmonicity. Note that as opposed to the transmon Depending on the relative magnitude of EJ and EL ,
qubit, the anharmonicity of the CSFQ is positive (α > 0). the fluxonium system could involve plasmon states (in
While the improvement in anharmonicity can be associ- the same well) and fluxon states (in different wells).
ated with reshaping the energy potential, the improved There are a variety of schemes to utilize them for quan-
coherence over the first flux qubits can be attributed to tum information processing. Generally, the spectrum of
 9

the transition between the lowest energy states is similar (a) Direct capacitive coupling
to that of the flux qubit, see Fig. 2(h). Both long coher- g12
ence and high anharmonicity can be expected at the flux
V1 V2
sweet spot. Cg
Lastly, we want to point out a further extension – the
0 − π qubit – which has even stronger topological protec- IC1 C1 IC2 C2
tion from noise83,84 . However, the strongly suppressed
sensitivity to external fluctuations also makes it hard to
manipulate. (b) Capacitive coupling via coupler
gr1 gr2

C. Interaction Hamiltonian engineering

Cg1 Cg2

To generate entanglement between individual quantum IC1 C1 Lr Cr IC2 C2

systems – it is necessary to engineer an interaction Hamil-
tonian that connects degrees of freedom in those indi-
vidual systems. In this section, we discuss the physical
coupling mechanism and its representation in the qubit (c) Direct inductive coupling

eigenbasis. The use of coupling to form 2-qubit gates is I1 M12 I2

discussed in Sec. IV.
IC1 L1 L2 IC2
Φe1 Φe2
1. Physical coupling: capacitive and inductive

The Hamiltonian of two coupled systems takes a

generic form (d) Inductive coupling via coupler
ICC
M1C M2C

H = H1 + H2 + Hint , (25) IC1 L1 L2 IC2

Φe1 ΦeC Φe2
where H1 and H2 denote the Hamiltonians of the individ-
ual quantum systems, which could be any combination of
the qubit circuits mentioned in Sec. II A and II B. The
FIG. 3. Schematic of capacitive and inductive coupling
last term, Hint , is the interaction Hamiltonian, which
schemes between two superconducting qubits, labeled 1 and
couples variables of both systems. In superconducting 2. (a) Direct capacitive coupling, where the voltage nodes
circuits, the physical form of the coupling energy is either of two qubits V1 and V2 are connected by a capacitance Cg .
an electric or magnetic field (or a combination thereof). (b) Capacitive coupling via a coupler in form of a linear res-
To achieve capacitive coupling, a capacitor is placed onator. (c) Direct inductive coupling, where the two qubits
between the voltage nodes of the two participating cir- are coupled via mutual inductance, M12 . (d) Inductive cou-
cuits, yielding an interaction Hamiltonian Hint of the pling via mutual inductances M1C and M2C to a frequency-
form tunable coupler.

Hint = Cg V1 V2 , (26)
where the expressions in brackets are the two Hamiltoni-
where Cg is the coupling capacitance and V1 (V2 ) is the ans of the individual qubits, [see Eq. (16)], and we take
voltage operator of the corresponding voltage node being Vi = (2e/Ci )ni in Eq. (26). From Eq. (27), we see that
connected. Fig. 3(a) illustrates a realistic example of a the coupling energy depends on the coupling capacitance
direct capacitive coupling between the top nodes of two as well as the matrix elements of the voltage operators.
transmon qubits. Circuit quantization in the limit of The dependencies are bilinear in the perturbative limit
Cg  C1 , C2 yields (Cg  C1 , C2 ).
To implement the coupling capacitance, one only need
bring the edges of the capacitor pads into close proxim-
X 
H= 4EC,i n2i − EJ,i cos(φi )

i=1,2 ity, as has been demonstrated in state-of-the-art planar
designs85 . The coupling capacitance is determined by
the planar capacitor geometry as well as the surround-
Cg ing environment, such as the dielectric constant of the
+4e2 n1 n2 , (27) substrate and the ground plane proximity.
C1 C2
 10

In the case of inductive coupling, a mutual inductance this simplifying approximation is only exact in the patho-
shared by two loops is the coupling mechanism, yielding logical limit of no coupling.
an interaction Hamiltonian that is of the intuitive form To realize a mutual inductance, two looped circuits
are brought into close proximity to one another, or, to
Hint = M12 I1 I2 , (28) make it stronger, overlap with each other88 , and even
may share the same wire or Josephson junction induc-
where M12 denotes the mutual inductance between the tor89–92 . In the case of a Josephson junction, and for
loops and I1 and I2 are the current operators for the certain metals, the inductance is dominated by kinetic
currents through the inductors. A typical example com- inductance contributions, rather than solely geometric
prises two closely positioned (rf-SQUID type) flux qubits, inductance93,94 . Kinetic inductance arises from the me-
as illustrated in Fig. 3(c). The system Hamiltonian can chanical, inertial mass of the charge carriers, but is only
be expressed as (see Refs. 86 and 87): practically witnessed in very high-conductance materi-
als like superconductors. A primary feature of kinetic
inductance is that its values can vastly exceed those of
X  1 Φ2Li

H= 4EC,i n2i − EJ,i cos(φi ) + conventional geometric inductances, which are generally
i=1,2
2 Li (1 − K 2 ) limited by electromagnetic considerations79 .
ΦL1 ΦL2
− M12 (1 − K 2 ) , (29)
L1 (1 − K 2 ) L2 (1 − K 2 ) 2. Coupling axis: transverse and longitudinal
where the first two terms are the energies associated with
the Josephson junctions, the third term captures the in- Regardless of its physical realization, the effect of a
ductor energies, and the fourth term is the mutual cou- coupling on system dynamics is determined by its form
pling energy of the form M I1 I2 (note that in general as represented in the eigenbasis of the individual systems.
Φ = LI). ΦL1 and ΦL2 are the magnetic fluxes asso- That is, how Hint appears in the representation spanned
ciated with the currents flowing through the respective by the eigenbasis of H1 ⊗ H2 .
inductors, and K is the
√ unitless mutual coupling factor Let us start with the previous example of two capaci-
defined by M12 = K L1 L2 . Importantly, note that in tively coupled transmon qubits [Fig. 3(a)]. Using second
Eq. (29), K 2 renormalizes L1 , L2 and M12 , essentially quantization, the system Hamiltonian in Eq. (27) can be
capturing the loading effect on the circuit due to their expressed as
presence, and is found by inverting an inductance matrix
(see Refs. 86 and 87). Xh † αi i
As we did with the charge degree of freedom, we will H= ωi ai ai + a†i a†i ai ai
2
normalize the inductor and externally applied magnetic i∈1,2

fluxes – in this case, by the reduced quantum unit of

flux Φ0 /2π – to define phases φ ≡ Φ/(Φ0 /2π) that are   
on the same footing as the junction phases φ1,2 . Due −g a1 − a†1 a2 − a†2 , (31)
to fluxoid quantization around the closed loop, these
phases must sum to zero or an integer multiple of 2π. where the expression within brackets represent the Duff-
For current directions entering the top of the junctions ing oscillator Hamiltonian for the qubits and g is the
[with I1 counterclockwise and I2 as shown in Fig. 3(c)], coupling energy. Since we define V ∝ n ∝ i(a − a† ), and
φi −φLi = φei +2πn with n = 0, ±1, . . . (see, for example, consequently I ∝ φ ∝ (a + a† ), the original n1 n2 -term
Ref. 72). Replacing φL1(L2) with φ1(2) − φ1(e2) yields: becomes what is shown in Eq. (31). Such a coupling is
called transverse, because the coupling Hamiltonian has
non-zero matrix elements only at off-diagonal positions
Φ0 2
" #
1 (φ )

2
X i − φ ei with respect to both oscillators, i.e. i hk|ai − a†i |kii = 0
= 4EC,i n2i − EJ,i cos(φi ) + 2π

i=1,2
2 Li (1 − K 2 ) for any integer k and for i ∈ 1, 2 and in this case
†
i hk ± 1|ai − ai |kii 6= 0.
Φ0
(φ1 − φe1 ) Φ2π (φ2 − φe2 )



0

− M12 (1 − K 2 ) 2π . (30) If we can ignore higher energy levels (k ≥ 2) either

L1 (1 − K )
2 L2 (1 − K 2 ) because of sufficient anharmonicity or through careful
control protocols that ensure these levels never have in-
The coupling is of the form M1,2 I1 I2 , with both the mu- fluence, we may truncate the Hamiltonian in Eq. (31)
tual coupling and the circulating currents (via the L1 to
and L2 ) “renormalized” by the factor (1 − K 2 ). To
capture the renormalization explicitly, the Hamiltonian
X 1
is generally simulated using phase operators (Eq. 30), H= ωi σz,i + gσy,1 σy,2 . (32)
rather than current operators. In the weak-coupling limit i∈1,2
2
K 2  1 (equivalently, M 2  L1 L2 ), the coupling term
may be approximated by the Josephson currents with This is a Hamiltonian of two spins, coupled by an ex-
M12 I1 I2 ≈ M12 Ic1 sin φ1 Ic2 sin φ2 . Note, however, that change interaction. As we will see in Sec. IV D 1, such
 11

a Hamiltonian is most commonly used in contemporary and approximation – be treated as an isolated system,
implementations and can generate various types of two- and the composite system simplified to two transversely
qubit entangling gates. Note that, more often, we see coupled qubits, see Eq. (32).
that the interaction term is expressed in σx σx instead of We now turn to the previous example of two induc-
σy σy . The choice in the context here is arbitrary and does tively coupled flux qubits, see Fig. 3(c). Assume that
not change the dynamics. However, when both capaci- the double-well potential [Fig. 2(g)] has a relatively
tive and inductive couplings are present in the system, high inter-well barrier, which leads to an exponentially
both σx σx and σy σy may be needed. In this case, the small qubit transition frequency at the energy degener-
voltage operator V ∝ i(a − a† ) (reduced to σy after two- acy point, (Φe = π). Around this degeneracy point, the
level approximation in the lab frame) is transversal to the off-diagonal matrix element of sin(φ) is zero, i.e. the
current operator I ∝ (a+a† ) (reduced to σx ) and both of ground and excited states are localized in different wells
them may be transverse to the qubit. A similar example and h1| sin(φ)|1i − h0| sin(φ)|0i =
6 0. We can then rewrite
is demonstrated between a qubit and a resonator by Lu the Hamiltonian in Eq. (??) as
et al.95
Transverse coupling can be engineered between a qubit X 1
and a harmonic oscillator, see Fig. 3(b). In this case, the H= ωi σzi + gσz1 σz2 . (35)
Hamiltonian becomes i=1,2
2

1 Now, the coupling axis is the same as the qubit quan-

H= ωq σz + ωr a† a + g(σ+ a + σ− a† ), (33) tization axes and therefore termed longitudinal coupling.
2
Note, however, that the physical σx σx and σz σz couplings
where ωq and ωr denote the qubit and resonator fre- can change in the qubit frame.
quencies, and σ+ = |0ih1| and σ− = |1ih0| describes Longitudinal coupling is an important type of inter-
the processes of exciting and de-exciting the qubit, re- action, because it can generate entanglement without
spectively. Here, we have assumed that the coupling is energy exchange. Moreover, it is found a necessary in-
in the dispersive limit, i.e. g  ωq , ωr , hence ignoring gredient in the application of quantum annealing, where
the double (de)excitation terms proportional to σ+ a† and certain hard combinatorial optimization problems can be
σ− a, which under typical operation regimes oscillate suf- modeled by the Ising Hamiltonian in Eq. (35) and finding
ficiently fast to average to zero. The Hamiltonian in Eq. its ground state would solve this problem.
(33), is the standard model used for describing how a two- An intermediate qubit mode may also be used as a cou-
level atom interacts with a resonant cavity that houses pler in the longitudinal case. In Fig. 3(d), an additional
it. Such a structure is also known as cavity quantum rf-SQUID is used to mediate the coupling. The coupling
electrodynamics (cQED), and it is extended to the cir- strength can be tuned by the flux bias of the coupler
cuit version here. It has many useful applications in su- SQUID103 . Note that a tunable coupler may also be re-
perconducting quantum information architectures, such alized in a structure with capacitive couplings63 . A tun-
as high-fidelity readout96 , see Sec. V, cavity buses97 , able coupler is useful because it provides a wide range of
quantum memory98,99 , quantum computation with cat coupling strengths87 , a high on-off ratio104 for reducing
states100–102 , etc. gate error-rates, and more ways of achieving high-fidelity
Here, we briefly mention the use of a cavity or res- entangling gates67,105–107 . The trade-off is an additional
onator to mediate coupling between two qubits, which control line.
may be physically well-separated (≈ 1 cm). Since most In addition to the pure transversal and longitudinal
superconducting resonators are in the GHz frequency qubit-qubit interactions, there are also examples of mixed
range, they can be made much longer than any dimen- types of interaction Hamiltonians108
sion of a qubit circuit (≈ 1 mm). One can use such a
resonator to mediate coupling between two or more oth-
erwise non-interacting qubits. An example is shown in 1
H= ωq σz + ωr a† a + gσz (a + a† ), (36)
Fig. 3(b), where two transmon qubits are both capac- 2
itively coupled to the center resonator. The two-level
system Hamiltonian is: which are longitudinal with respect to a qubit, but trans-
verse with respect to a harmonic oscillator in a qubit-
resonator system. Such a model is called longitudinal
X  αi † † 
but one should note that it is only longitudinal to one
H= ωi a†i ai + ai ai ai ai + ωr a†r ar
i=1,2
2 participating system. It is hard to engineer physically
longitudinal coupling with respect to a harmonic oscil-
    lator, since either the E-field (V ) or the B-field (I) is
+g1r a†1 ar + a1 a†r + g2r a†2 ar + a2 a†r . (34) transverse with respect to the eigen field of the harmonic
oscillator. Note, however, that a transversal model such
It can be shown that in the dispersive limit, i.e. gir  as in Eq. (33) may be transformed into a longitudinal
|ωi − ωr |, the resonator can – after proper transformation one in certain operating regimes, see Sec. V.
 12

In some applications, such as for quantum annealing, 2. Stochastic noise

both longitudinal and transverse couplings are desired
(σz σz coupling for mapping the problem and σx σx cou- The second type of noise is stochastic noise, arising
pling for enhancing the annealing performance) and re- from random fluctuations of parameters that are coupled
quire independent control. to our qubit109 . For example, thermal noise of a 50Ω re-
sistor in the control lines leading to the qubit will have
voltage and current fluctuations – Johnson noise – with a
III. NOISE, DECOHERENCE, AND ERROR noise power that is proportional to both temperature and
MITIGATION
bandwidth. Or, the oscillator that provides the carrier
for a qubit control pulse may have amplitude or phase
Random, uncontrollable physical processes in the qubit fluctuations. Additionally, randomly fluctuating electric
control and measurement equipment or in the local envi- and magnetic fields in the local qubit environment – e.g.,
ronment surrounding the quantum processor are sources on the metal surface, on the substrate surface, at the
of noise that lead to decoherence and reduce the opera- metal-substrate interface, or inside the substrate – can
tional fidelity of the qubits. In this section, we introduce couple to the qubit. This creates unknown and uncon-
the basics of noise leading to decoherence in supercon- trolled fluctuations of one or more qubit parameters, and
ducting circuits, and we discuss coherent control methods this leads to qubit decoherence.
to mitigate certain types of noise.

3. Noise strength and qubit susceptibility

A. Types of noise

The degree to which a qubit is affected by noise is

In a closed system, the dynamical evolution of a qubit
related to the amount of noise impinging on the qubit,
state is deterministic. That is, if we know the starting
and the qubit’s susceptibility to that noise. The former
state of the qubit and its Hamiltonian, then we can pre-
is often a question of materials science and fabrication;
dict the state of the qubit at any time in the future.
that is, can we make devices with lower levels of noise.
However, in open systems, the situation changes. The
Or, it may be related to the quality of the control elec-
qubit now interacts with uncontrolled degrees of freedom
tronics and cryogenic engineering to limit the levels of
in its environment, which we refer to as fluctuations or
noise on the control lines that necessarily connect to the
noise. In the presence of noise, as time progresses, the
qubits to control them. The latter – qubit susceptibility
qubit state looks less and less like the state we would
– is a question of qubit design. Qubits can be designed to
have predicted and, eventually, the state is lost. There
trade off sensitivity to one type of noise at the expense of
are many different sources of noise that affect quantum
increased sensitivity to other types of noise. Thus, mate-
systems, and they can be categorized into two primary
rials science, fabrication engineering, electronics design,
types: systematic noise and stochastic noise.
cryogenic engineering, and qubit design all play a role
in creating devices with high coherence. In general, one
should strive to eliminate the sources of noise, and then
1. Systematic noise
design qubits that are insensitive to the residual noise.
The qubit response to noise depends on how the noise
Systematic noise arises from a process that is trace- couples to it – either through a longitudinal or a trans-
able to a fixed control or readout error. For example, we verse coupling as referenced to the qubit quantization
apply a microwave pulse to the qubit that we believe axis. This can be visualized using a Bloch Sphere picture
will impart a 180-degree rotation. However, the con- of the qubit state, as illustrated in Fig. 4 and discussed
trol field is not tuned properly and, rather than rotating in detail in Section III B.
the qubit 180 degrees, the pulse slightly over-rotates or
under-rotates the qubit by a fixed amount. The underly-
ing error is systematic, and it therefore leads to the same
B. Modeling noise and decoherence
rotation error each time it is applied. However, when
such erroneous pulses are used in practice in a variety
of control sequences, the observed results may appear 1. Bloch sphere representation
to be influenced by random noise. This is because the
pulse is generally not applied in the same way for each The Bloch sphere is a unit sphere used to represent the
experiment: it could be applied a different number of quantum state of a two-level system (qubit). Fig. 4(a)
times, interspersed with different pulses in different or- shows a Bloch sphere with a Bloch vector representing
ders, and therefore generally differs from experiment to the state |ψi = α|0i + β|1i. If we visualize the Bloch
experiment. However, once systematic errors are iden- sphere as the planet Earth, then by convention, the north
tified, they can generally be corrected through proper pole represents state |0i and the south pole state |1i. For
calibration or the use of improved hardware. pure quantum states such as |ψi, the Bloch vector is of
 13

(a) Bloch sphere (b) Longitudinal relaxation (c) Pure dephasing (d) Transverse relaxation
Longitudinal
z (Longitudinal) z z noise
z
|0〉 |0〉 |0〉 |0〉
|Ψ〉 = α|0〉 + β|1〉 Pure
Excitation Relaxation Dephasing Decoherence
θ Γ1 Γ1 Γφ Γ2
Γ1
y y y y
φ (Transverse)
x x Transverse x Γφ x Γφ
(Transverse) noise
Transverse
|1〉 noise |1〉 |1〉 |1〉

FIG. 4. Transverse and longitudinal noise represented on the Bloch sphere. (a) Bloch sphere representation of the quantum
state |ψi = α|0i + β|1i. The qubit quantization axis – the z axis – is longitudinal in the qubit frame, corresponding to σz
terms in the qubit Hamiltonian. The x-y plane is transverse in the qubit frame, corresponding to σx and σy terms in the
qubit Hamiltonian. (b) Longitudinal relaxation results from energy exchange between the qubit and its environment, due
to transverse noise that couples to the qubit in the x − y plane and drives transitions |0i ↔ |1i. A qubit in state |1i emits
energy to the environment and relaxes to |0i with a rate Γ1↓ (blue arched arrow). Similarly, a qubit in state |0i absorbs energy
from the environment, exciting it to |1i with a rate Γ1↑ (orange arched arrow). In the typical operating regime kB T  ~ωq ,
the up-rate is suppressed, leading to the overall decay rate Γ1 ≈ Γ1↓ . (c) Pure dephasing in the transverse plane arises from
longitudinal noise along the z axis that fluctuates the qubit frequency. A Bloch vector along the x-axis will diffuse clockwise or
counterclockwise around the equator due to the stochastic frequency fluctuations, depolarizing the azimuthal phase with a rate
Γφ . (d) Transverse relaxation results in a loss of coherence at a rate Γ2 = Γ1 /2 + Γφ , √
due to a combination of energy relaxation
and pure dephasing. Pure dephasing leads to decoherence of the quantum state (1/ 2)(|0i + |1i), initially pointed along the
x-axis. Additionally, the excited state component of the superposition state may relax to the ground state, a phase-breaking
process that loses the orientation of the vector in the x-y plane.

unit length, |α|2 + |β|2 = 1, connecting the center of the |ψihψ| for a pure state |ψi is equivalently
sphere to any point on its surface.
The z-axis connects the north and south poles. It 1 1 1 + cos θ e−iφ sin θ
 
is called the longitudinal axis, since it represents the ρ= (I + ~a · ~σ ) = (38)
qubit quantization axis for the states |0i and |1i in the 2 2 eiφ sin θ 1 + sin θ
qubit eigenbasis. In turn, the x-y plane is the trans- cos2 θ2 e−iφ cos θ2 sin θ2
 
= iφ
verse plane with transverse axes x and y. In this e cos θ2 sin θ2 sin2 θ2
Cartesian coordinate system, the unit Bloch vector ~a = (39)
(sin θ cos φ, sin θ sin φ, cos θ) is represented using the polar  2 ∗

|α| αβ
angle 0 ≤ θ ≤ π and the azimuthal angle 0 ≤ φ < 2π, as = (40)
α∗ β |β|2
illustrated in Fig. 4 (a). Following our convention, state
|0i at the north pole is associated with +1, and state |1i
(the south pole) with −1. We can similarly represent the
where I is the identity matrix, and ~σ = [σx , σy , σz ] is a
quantum state using the angles θ and φ,
vector of Pauli matrices. If the Bloch vector ~a is a unit
vector, then ρ represents a pure state ψ and Tr(ρ2 ) = 1.
θ θ More generally, the Bloch sphere can be used to represent
|ψi = α|0i + β|1i = cos |0i + eiφ sin |1i. (37) mixed states, for which |~a| < 1; in this case, the Bloch
2 2
vector terminates at points inside the unit sphere, and
0 ≤ Tr(ρ2 ) < 1. To summarize, the surface of the unit
The Bloch vector is stationary on the Bloch sphere in sphere represents pure states, and its interior represents
the rotating frame picture. If state |1i has a higher en- mixed states6 .
ergy than state |0i (as it generally does in superconduct-
ing qubits), then in a stationary frame, the Bloch vector
would precess around the z-axis at the qubit frequency
(E1 −E0 )/~. Without loss of generality (and much easier
to visualize), we instead choose to view the Bloch sphere
in a reference frame where the x and y-axes also rotate 2. Bloch-Redfield model of decoherence
around the z-axis at the qubit frequency. In this rotating
frame, the Bloch vector appears stationary as written in
Within the standard Bloch-Redfield110–112 picture of
Eq. (37). The rotating frame will be described in detail
two-level system dynamics, noise sources weakly coupled
in Section IV D 1 in the context of single-qubit gates.
to the qubits have short correlation times with respect
For completeness, we note that the density matrix ρ = to the system dynamics. In this case, the relaxation pro-
 14

cesses are characterized by two rates (see Fig. 4): ground state (|0i) at the north pole, p = −1 is entirely
in the excited state (|1i) at the south pole, and p = 0 is
1 a completely depolarized mixed state at the center of the
longitudinal relaxation rate: Γ1 ≡ (41)
T1 Bloch sphere.
1 Γ1 As illustrated in Fig. 4(b), longitudinal relaxation is
transverse relaxation rate: Γ2 ≡ = + Γϕ
T2 2 caused by transverse noise, via the x- or y-axis, with
(42) the intuition that off-diagonal elements of an interaction
Hamiltonian are needed to connect and drive transitions
which contains the pure dephasing rate Γϕ . We note between states |0i and |1i.
that the definition of Γ2 as a sum of rates presumes that Depolarization occurs due to energy exchange with an
the individual decay functions are exponential, which oc- environment, generally leading to both an “up transition
curs for Lorentzian noise spectra (centered at ω = 0) rate” Γ1↑ (excitation from |0i to |1i), and a “down tran-
such as white noise (short correlation times) with a high- sition rate” Γ1↓ (relaxation from |1i to |0i). Together,
frequency cutoff. these form the longitudinal relaxation rate Γ1 :
The impact of noise on the qubit can be visualized on
the Bloch sphere in Fig. 4(a). For an initial state (t = 0) 1
Γ1 ≡ = Γ1↓ + Γ1↑ . (45)
T1
|ψi = α|0i + β|1i, (43)
T1 is the 1/e decay time in the exponential decay func-
the Bloch-Redfield density matrix ρBR for the qubit is tion in Eq. (44), and it is the characteristic time scale
written113,114 , over which qubit population will relax to its steady-
state value. For superconducting qubits, this steady-
1 + (|α|2 − 1)e−Γ1 t αβ ∗ eiδωt e−Γ2 t
 
ρBR = . (44) state value is generally the ground state, due to Boltz-
α∗ βe−iδωt e−Γ2 t |β|2 e−Γ1 t mann statistics and typical operating conditions. Boltz-
mann equilibrium statistics lead to the “detailed bal-
There are a few important distinctions between Eq. (44) ance” relationship Γ1↑ = exp(−~ωq /kB T )Γ1↓ , where T
and Eq. (40), which we list here and then describe in is the temperature and kB is the Boltzmann constant,
more detail in subsequent sections. with an equilibrium qubit polarization approaching p =
• First, we have introduced the longitudinal decay tanh(~ωq /2kB T ). Typical qubits are designed at fre-
function exp(−Γ1 t), which accounts for longitudi- quency ωq /2π ≈ 5 GHz and are operated at dilution
nal relaxation of the qubit. refrigerator temperatures T ≈ 20 mK. In this limit, the
up-rate Γ1↑ is exponentially suppressed by the Boltz-
• Second, we introduced the transverse decay func- mann factor exp(−~ωq /kB T ), and so only the down-
tion exp(−Γ2 t), which accounts for transverse de- rate Γ1↓ contributes significantly, relaxing the popula-
cay of the qubit. tion to the ground state. Thus, qubits generally spon-
taneously lose energy to their cold environment, but the
• Third, we have introduced an explicit phase ac- environment rarely introduces a qubit excitation. As a
crual exp(iδωt), where δω = ωq − ωd , which gen- result, the equilibrium polarization approaches unity [see
eralizes the Bloch sphere picture to account for Eq. (44)]118,119 .
cases where the qubit frequency ωq differs from the Only noise at the qubit frequency mediates qubit tran-
rotating-frame frequency ωd , as we will see later sitions, whether absorption or emission, and this noise is
when discussing measurements of T2 using Ramsey generally “well behaved” (short correlation time, many
interferometry115,116 , and in Section IV D 1 in the modes weakly coupled to qubit, no divergences) around
context of single-qubit gates. the qubit frequency for superconducting qubits. The in-
• And, fourth, we have constructed the matrix such tuition is that qubit-transition linewidths are relatively
that for t  (T1 , T2 ), the upper-left matrix ele- narrow in frequency, and so the noise generally does not
ment will approach unit value, indicating that all vary much over this narrow frequency range. Although
population relaxes to the ground state, while the there are a few notable exceptions, for example, qubit
other three matrix elements decay to zero. This is decay in the presence of hot quasiparticles120–122 , which
related to the assumption that the environmental can lead to non-exponential decay functions, longitudinal
temperature is low enough that thermal excitations depolarization measurements generally present exponen-
of the qubit from the ground to excited state rarely tial decay functions consistent with the Bloch-Redfield
occur. picture.
An example of a T1 measurement is shown in Fig. 5(a).
Longitudinal relaxation The qubit is prepared in its excited state using an Xπ -
The longitudinal relaxation rate Γ1 describes depolar- pulse, and then left to spontaneously decay to the ground
ization along the qubit quantization axis, often referred state for a time τ , after which the qubit is measured. A
to as “energy decay” or “energy relaxation.” In this lan- single measurement will project the quantum state into
guage, a qubit with polarization p = 1 is entirely in the either state |0i or state |1i, with probabilities that cor-
 15

(a) (c)

Xπ Readout Xπ
X π/2 X π/2
Relaxation: τ Echo: τ/2 τ/2
t t

(b) (d)

X π/2 X π/2
Ramsey:
τ
t

FIG. 5. Characterizing longitudinal (T1 ) and transverse (T2 ) relaxation times of a transmon qubit117 . (a) Longitudinal
relaxation (energy relaxation) measurement. The qubit is prepared in the excited state using an Xπ -pulse and measured after
a waiting time τ . For each value τ , this procedure is repeated to obtain an ensemble average of the qubit polarization: +1
corresponding to |0i, and −1 corresponding to |1i. The resulting exponential decay function has a characteristic time T1 = 85
µs. (b) Transverse relaxation (decoherence) measurement via Ramsey interferometry. The qubit is prepared on the equator
using an Xπ/2 -pulse, intentionally detuned from the qubit frequency by δω, causing the Bloch vector to precess in the rotating
frame at a rate δω around the z-axis. After a time τ , a second Xπ/2 pulse then projects the Bloch vector back on to the
z axis, effectively mapping its former position on the equator to a position on the z axis. The oscillations decay with an
approximately (but not exactly) exponential decay function, with a characteristic time T2∗ = 95 µs. (c) Transverse relaxation
(decoherence) measurement via a Hahn echo experiment116 . The qubit is prepared and measured in the same manner as the
Ramsey interfometry experiment, except that a single Xπ pulse is applied midway through the free-evolution time τ . The
decay function is approximately exponential, with a characteristic time T2E = 120 µs. The coherence improvement using the
Hahn echo over panel (b) indicates that some low-frequency dephasing noise has been mitigated; however, a small amount
remains since T2E has not yet reached the 2T1 limit. (d) Coherence function incorporating T1 loss and Gaussian dephasing
components of the Ramsey interferometry data in panel (b). The Gaussian-distributed 1/f noise spectrum of magnetic flux
noise leads to a decay function exp(−t/2T1 ) exp(−χN ) = exp(−t/2T1 ) exp(−t2 /Tϕ,G
2
) in Eq. (46). These two decay functions
together match well the Ramsey data in panel (b).

respond to the qubit polarization. To make an estimate in the x − y plane of the Bloch sphere. It is referred to
of this polarization, one needs to identically prepare the as “pure dephasing,” to distinguish it from other phase-
qubit and repeat the experiment many times. This is breaking processes such as energy excitation or relax-
analogous to flipping a coin: any single flip will yield ation.
heads or tails, but the probability of obtaining a heads
or tails can be estimated by flipping the coin many times As illustrated in Fig. 4(c), pure dephasing is caused
and taking the ensemble average. The resulting expo- by longitudinal noise that couples to the qubit via the z-
nential decay has a characteristic time T1 = 85 µs. axis. Such longitudinal noise causes the qubit frequency
ωq to fluctuate, such that it is no longer equal to the
Pure dephasing
rotating frame frequency ωd , and causes the Bloch vec-
The pure dephasing rate Γφ describes depolarization tor to precess forward or backward in the rotating frame.
 16

Intuitively, we can imagine identically preparing several Bloch vector back on to the z-axis. Repeated measure-
instances of the Bloch vector along the x-axis. For each ments are made to take an ensemble averaged estimate of
instance, the stochastic fluctuations of qubit frequency the qubit polarization, as a function of τ . The resulting
will result in a different precession frequency, resulting oscillations in Fig. 5(b) feature an approximately expo-
in a net fanout of the Bloch vector in the x − y plane. nential decay function with time T2∗ = 98 µs. The “*”
This eventually leads to a complete depolarization of the indicates that the Ramsey experiment is sensitive to in-
azimuthal angle φ. Note that this stochastic effect will be homogeneous broadening. That is, it is highly sensitive
captured in the transverse relaxation rate Γ2 (next sec- to quasi-static, low-frequency fluctuations that are con-
tion); it is not the deterministic term exp(±iδωt) that stant within one experimental trial, but vary from trial
appears in Eq. (44), which represents intentional detun- to trial, e.g., due to 1/f -type noise. This sensitivity to
ing of the qubit reference frame. quasi-static noise is related to the corresponding N = 0
There are a few important distinctions between pure noise filter function shown in Fig. 5(d) being centered
dephasing and energy relaxation. First, in contrast to at at zero-frequency, as described in more detail in Sec-
energy relaxation, pure dephasing is not a resonant phe- tion III D 2.
nomenon; noise at any frequency can modify the qubit The Hahn echo shown in Fig. 5(c) is an experiment
frequency and cause dephasing. Thus, qubit dephasing that is less sensitive to quasi-static noise. By placing
is subject to broadband noise. Second, since pure de- a Yπ pulse at the center of a Ramsey interferometry ex-
phasing is elastic (there is no energy exchange with the periment, the quasi-static contributions to dephasing can
environment), it is in principle reversible. That is, the be “refocused,” leaving an estimate T2E that is less sen-
dephasing can be “undone” – with quantum informa- sitive to inhomogeneous broadening mechanisms. The
tion being preserved – through the application of unitary pulses are generally chosen to be resonant with the qubit
operations, e.g., dynamical decoupling pulses78 , see Sec. transition for a Hahn echo, since any frequency detuning
III D 2. would be nominally refocused anyway. The resulting de-
The degree to which the quantum information can be cay function in Fig. 5(c) is essentially exponential with
retained depends on many factors, including the band- time T2E = 120 µs.
width of the noise, the rate of dephasing, the rate at With the known T1 and T2 times, one can infer the pure
which unitary operations can be performed, etc. This dephasing time Tϕ from Eq. (42), provided the decay
should be contrasted with spontaneous energy relaxation, functions are exponential. In superconducting qubits,
which is an irreversible process. Intuitively, once the however, the broadband dephasing noise (e.g., flux noise,
qubit emits energy to the environment and its myriad charge noise, critical-current noise, ...) tends to exhibit
uncontrollable modes, the quantum information is essen- a 1/f -like power spectrum. Such noise is singular near
tially lost with no hope for its recovery and reconstitution ω = 0, has long correlation times, and generally does
back into the qubit. not fall within the Bloch-Redfield description. The de-
Transverse relaxation cay function of the off-diagonal terms in Eq. (44) are
The transverse relaxation rate Γ2 = Γ1 /2 + Γϕ describes generally non-exponential, and for such cases, the simple
the√ loss of coherence of a superposition state, for example expression in Eq. (42) is not applicable.
(1/ 2)(|0i+|1i), pointed along the x-axis on the equator
of the Bloch sphere as illustrated in Fig. 4(d). Decoher-
ence is caused in part by longitudinal noise, which fluctu- 3. Modification due to 1/f -type noise
ates the qubit frequency and leads to pure dephasing Γϕ
(red). It is also caused by transverse noise, which leads
If we assume that the qubit is coupled to many in-
to energy relaxation of the excited-state component of
dependent fluctuators, then, regardless of their individ-
the superposition state at a rate Γ1 (blue). Such a relax-
ual statistics, they will in concert generate noise with
ation event is also a phase-breaking process, because once
a Gaussian distribution due to the central limit theo-
it occurs, the Bloch vector points to the north pole, |0i,
rem. We therefore say that the longitudinal fluctuations
and there is no longer any knowledge of which direction
exhibit Gaussian-distributed 1/f noise123,124 . For 1/f
the Bloch vector had been pointing along the equator;
noise spectra, the phase decay function is itself a Gaus-
the relative phase of the superposition state is lost.
sian exp −(t/Tϕ,G )2 , where we write Tϕ,G to distinguish
 
Transverse relaxation T2 can be measured using Ram- it from Tϕ used in Eq. (42). Furthermore, this function
sey interferometry, as shown and described in Fig. 5(b). is separable from the T1 -type exponential decay, because
The protocol positions the Bloch vector on the equator the T1 -noise remains regular at the qubit frequency. The
using a Xπ/2 -pulse. Typically, the carrier frequency of density matrix in Eq. (44) becomes, following Refs. 78
this pulse is slightly detuned from the qubit frequency and 113,
by an amount δω. As a result, the Bloch vector will pre-
cess around the z-axis at a rate δω. This is done for Γ1
!
convenience sake, so that the resulting Ramsey measure- 1 + (|α|2 − 1)e−Γ1 t αβ ∗ eiδωt e− 2 t e−χN (t)
ρ= Γ1 ,
ment will oscillate, making it easier to analyze. After α∗ βe−iδωt e− 2 t e−χN (t) |β|2 e−Γ1 t
precessing for a time τ , a second Xπ/2 -pulse projects the (46)
 17

For completeness, in addition to 1/f dephasing mech-

anisms, we note that there are also “white” pure dephas-
Qubit absorption Qubit emission ing mechanisms, which give rise to an exponential de-
from environment to environment
cay function for the dephasing component of T2 . One
common example is dephasing due to the shot noise of
Thermal 1/f Johnson- residual photons in the readout resonator coupled to su-
(Johnson) noise Nyquist
noise Quantum perconducting qubits, as we discuss in Section III C 3.
noise (Nyquist)
noise
4. Noise power spectral density (PSD)

The frequency distribution of the noise power for a sta-

tionary noise source λ is characterized by its PSD Sλ (ω)
Z ∞
“Low-frequency” Sλ (ω) = dτ hλ(τ )λ(0)ie−iωτ . (47)
noise −∞

The Wiener-Khintchine theorem states that the PSD

FIG. 6. Examples of symmetric and asymmetric noise spec-
tral densities. Noise at positive (negative) frequencies cor- is the Fourier transform of the autocorrelation function
responds to the qubit emitting (absorbing) energy to (from) cλ (τ ) = hλ(τ )λ(0)i of the noise source λ. Since the in-
its environment. Thermal noise is proportional to temper- tegration limits are (−∞, ∞), this is the bilateral PSD.
ature T and carries essentially a white noise spectrum. As Symmetrizing the PSD allows one to consider only posi-
it represents a classical fluctuating parameter, such as elec- tive frequencies, which is termed a unilateral PSD. Both
tric current, the noise power spectral density is symmetric unilateral and bilateral PSDs are used, often with the
in frequency. When resonant with the qubit, it will drive same notation, and so one needs to know how the PSD
both stimulated emission and absorption processes. The qubit is defined, keep track of the factors of 2 and π, and also
may also spontaneously emit energy to its environment, rep- be aware of the implications for quantum systems.
resented as Nyquist noise125 , a quantum mechanical effect
For classical systems, the noise power spectral density
that is not symmetric in frequency. At sufficiently low tem-
peratures or high frequencies, ~ω > 2kB T , the Nyquist noise
is symmetric. This is because the autocorrelation func-
dominates thermal noise. Another common example is 1/f tion of real signals is itself a real function, and the Fourier
noise, which is also a classical noise fluctuation and symmet- transform of a real temporal function is symmetric in the
ric in frequency. frequency domain. Dephasing noise is caused by real,
fluctuating fields, and so its PSD is generally symmetric.
Examples of such classical noise include thermal (John-
son) noise and 1/f noise126 (see Fig. 6).
where the decay function hexp(−χN (t))i contains the co-
In turn, the inverse Fourier transform of the PSD will
herence function χN (t), which generalizes pure dephas-
yield the autocorrelation function:
ing to include non-exponential decay functions. As we
shall see later, the subscript N labeling the decay func- 1
Z ∞
tion refers to the number of π-pulses used to refocus the cλ (τ ) = dω Sλ (ω)eiωτ . (48)
2π −∞
low-frequency noise, which impacts the form of the de-
cay function. Because the function is no longer purely This implies that integrating the noise power spectral
exponential, we cannot formally write the transverse re- density with τ = 0 yields the second moment of the noise,
laxation decay function as exp(−t/T2 ). However, an ex- or, for zero-mean fluctuations, the variance.
ponential decay remains a practically reasonable approx- However, the autocorrelation function for a quantum
imation for Tϕ & T1 . We also note that the energy decay system may be complex-valued due to the fact that quan-
component of the transverse relaxation is exp(−t/2T1 ), tum operators generally do not commute at different
and so T2 can never be larger than 2T1 . In the absence times. This means that time-ordering of the operators
of pure dephasing, the maximum T2 = 2T1 is reached. matters, and the PSD need not be symmetric in fre-
As an example, consider the Ramsey interferometry quency. This is generally the case for transverse noise
data in Fig. 5(b). Since the dephasing is relatively weak, causing longitudinal energy relaxation. Noise at a posi-
the transverse relaxation function as exp(−t/T2 ) is a tive frequency S(ωq ) corresponds to energy transfer from
reasonable fit and yields T2 = 95 µs. However, us- the qubit to the environment, including both stimulated
ing the value T1 = 85 µs from Fig. 5(a) and dividing and spontaneous emission, associated with the down-rate
out exp(−t/2T1 ) from the data in Fig. 5(b), the remain- Γ1↓ . Noise at a negative frequency S(−ωq ) corresponds
ing pure dephasing decay function is shown in Fig. 5(d) to energy transfer to the qubit from the environment, as-
and  assumes a Gaussian envelope hexp(−χN (t))i = sociated with the up-rate Γ1↑ . For a detailed discussion,
exp −(t/Tϕ,G t)2 , with Tϕ,G = 98 µs. The Hahn echo see Refs. 127 and 128. Spontaneous emission to a cold
data in Fig. 5(c) may be treated similarly. environment or electromagnetic vacuum, represented by
 18

Nyquist noise in Fig. 6, is an example of an asymmetric SQ (ω) = BQ 2

[ω/(2π ×1Hz)], where the noise strength BQ 2

noise PSD125 . at 1 Hz can assume a range of values depending on the

In general, making a connection between Sλ (ω) and level of dissipation in the system. Likewise, the cross-over
the measured qubit decay functions is the basis for noise from 1/f -like behavior to f -like behavior generally occurs
spectroscopy up to second-order statistics78,129–132 . The at around 1 GHz, but will vary higher or lower between
search for higher-order spectra related to non-Gaussian samples depending on the degree of dissipation62,136 .
noise is a current topic of active research133 .

C. Common examples of noise

2. Magnetic flux noise
There are many sources of stochastic noise in super-
conducting qubits, and we refer the reader to Ref. 40 for Another commonly observed noise in solid-state de-
a review. Here, we briefly present several of the most vices is magnetic flux noise. The origin of this noise is
common types of noise, their affect on coherence, and understood to arise from the stochastic flipping of spins
refer the reader to the references for a more detailed dis- (magnetic dipoles) that reside on the surfaces of the su-
cussion. perconducting metals comprising the qubit137 , resulting
in random fluctuations of the effective magnetic field that
biases flux-tunable qubits.
1. Charge noise
For example, in the case of the split transmon, the
external magnetic field threading the loop couples longi-
Charge noise is ubiquitous in solid-state devices. It tudinally to the qubit and modulates the transition fre-
arises from charged fluctuators present in the defects or quency via the Josephson energy EJ (except at ϕe = 0,
charge traps that reside in interfacial dielectrics, the junc- where the qubit is first-order insensitive to magnetic-
tion tunnel barrier, and in the substrate itself. These field fluctuations). Because the flux noise is longitudinal
are often modeled as an ensemble of fluctuating two-level to the transmon, it contributes to pure dephasing (Tϕ ).
systems or as bulk dielectric loss134,135 . For example, in However, in the case of the flux qubit, and depending on
the case of a transmon qubit, the electric field between the flux-bias point, the flux noise may be either longitu-
the capacitor plates traverses and couples to dielectric dinal – causing dephasing Tϕ – or it may couple trans-
defects residing on the metal surfaces of the plates (for versely and thus contribute to T1 relaxation62,78 . The
lateral-plate-type capacitors) or the capacitor dielectric noise power spectrum of these fluctuations generally ex-
between the plates (for parallel-plate-type capacitors). hibits a “quasi-universal” dependence,
The electric field variable is transverse with respect to
the quantization axis of the transmon qubit, which means γΦ
2π × 1Hz

that this noise is mainly responsible for energy relaxation SΦ (ω) = A2Φ , (50)
(T1 ). Additionally, if the EJ /EC ratio of the transmon is ω
not made sufficiently large (smaller than around 60), the
qubit frequency itself will also be sensitive to broadband with γΦ ≈ 0.8 − 1.0 and A2Φ ≈ (1 µΦ0 )2 /Hz, and has
charge fluctuations. In this case, low-frequency charge been shown to extend from less than millihertz to beyond
noise couples longitudinally to the transmon and causes gigahertz frequencies78,131,132,138,139 .
pure dephasing (Tϕ ).
Charge noise is modeled primarily as a combination of The large, low-frequency weighting of the 1/f power
inverse-frequency noise and Nyquist noise, also referred distribution enables the use of engineered error miti-
to as ohmic noise. At lower frequencies, the spectral gation techniques – such as dynamical decoupling – to
density takes the form achieve better coherence78,140–142 and for improving sin-
gle and two-qubit gate fidelity143 . It was recently demon-
strated that 1/f flux noise is also a T1 -mechanism when
γ
2π × 1Hz Q

SQ (ω) = A2Q , (49) extended out to the qubit frequency62 , and one similarly
ω
expects a crossover to ohmic flux noise at high enough
with quasi-universal values A2Q = (10−3 e)2 /Hz at 1 Hz, frequencies144 .
and γQ ≈ 1. In addition to large 1/f fluctuations, early Although much is known about the statistics and
charge qubits often witnessed discrete, charge offsets rem- number of the defects presumed responsible for flux
iniscent of random telegraph noise. Together, these two noise, their precise physical manifestation remains uncer-
mechanisms severely limited the utility of charge qubits, tain137,145 . The fact that the 1/f noise is quasi-universal
and served as a strong motivation to move to capaci- and largely independent of device, strongly suggests a
tively shunted charge qubits (transmons), which greatly common origin for the noise. Recent studies suggest that
reduced the qubit longitudinal sensitivity to charge noise. adsorbed molecular oxygen may be responsible for flux-
At higher frequencies, the power spectrum takes the form noise145,146 .
 19

3. Photon number fluctuations pair149,150 . Although this quasiparticle generation mech-

anism is not yet well understood, it has been shown that
In the circuit QED architecture, resonator pho- quasiparticles can be transiently pumped away, improv-
ton number fluctuation is another major decoherence ing T1 times and reducing T1 temporal variation122 .
source147 . Residual microwave fields in the cavity have
photon-number fluctuations that in the dispersive regime
impact the qubit through an interaction term χσz n, see D. Operator form of qubit-environment interaction
Sec. II C 2, leading to a frequency shift ∆Stark = 2ηχn̄,
where n̄ is the average photon number, and η = κ2 /(κ2 + Similar to the way that two qubits are coupled, a qubit
4χ2 ) effectively scales the photon population seen by the may couple and interact with uncontrolled degrees of
qubit due to the interplay between the qubit-induced dis- freedom (DOF) in its environment (the noise sources).
persive shift of the resonator frequency (χ) and the res- The interaction Hamiltonian between the qubit DOF
onator decay rate (κ). (Ôq ) and those of the noise source (λ̂) may be expressed
In the dispersive limit, the noise is longitudinally cou- in a general form
pled to the qubit and leads to pure dephasing at a rate,
4χ2 Ĥint = ν Ôq λ̂ (53)
Γφ = η n̄. (51)
κ
where ν denotes the coupling strength – which is related
The fluctuations originate from residual photons in the to the sensitivity of the qubit to environmental fluctua-
resonator, typically due to radiation from higher tem- tions ∂ Ĥq /∂λ – and we assume that Ôq is a qubit opera-
perature stages in the dilution refrigerator107,148 . The
tor within the qubit Hamiltonian Ĥq . The noisy environ-
corresponding noise spectral density is of a Lorentzian
type, ment represented by the operator λ̂ produces fluctuations
δλ. Note that we retained the hats in this section to re-
2ηn̄κ mind us that these are quantum operators.
S(ω) = 4χ2 , (52)
ω 2 + κ2
which exhibits an essentially white noise spectrum up to 1. Connecting T1 to S(ω)
a 3dB cutoff frequency ω = κ set by the resonator decay
rate κ, see Ref. 62.
If the coupling is transverse to the qubit, e.g. Ôq
is of the type σx or (a + a† ) – see the related case of
4. Quasiparticles qubit-qubit coupling treated in Sec. II C – then noise
at the qubit frequency can cause transitions between
Quasiparticles, i.e. unpaired electrons, are another im- the qubit eigenstates. Since this is a stochastic process,
portant noise source for superconducting devices121 . The the ensemble-average manifests itself as a decay (usually
tunneling of quasiparticles through a qubit junction may exponential) of the qubit population towards a certain
lead to both T1 relaxation and pure dephasing Tϕ , de- equilibrium value (usually the qubit ground state |0i for
pending on the type of qubit, the bias point, and the kB T  ~ωq ). Again, this process is equivalently referred
junction through which the tunneling event occurs120,122 . to as “T1 relaxation”, “energy relaxation”, or “longitudi-
Quasiparticles are naturally excited due to thermo- nal relaxation”. As stated above, T1 is the characteristic
dynamics, and the quasiparticle density in equilibrium time scale of the decay. Its inverse, Γ1 = 1/T1 is called
superconductors should be exponentially suppressed as the relaxation rate and depends on the power spectral
temperature decreases. However, below about 150 mK, density of the noise S(ω) at the transition frequency of
the quasiparticle density observed in superconducting de- the qubit ω = ωq :
vices – generally in the range 10−8 −10−6 per Cooper pair
– is much higher than BCS theory would predict for a su-  2
perconductor in equilibrium with its cryogenic environ- 1  ∂ Ĥ
q

Γ1 = 2 h0| |1i Sλ (ωq ), (54)
 
ment at 10 mK. The reason for this excess quasiparticle ~  ∂λ 
population is unclear, but it is very likely related to the
presence of additional, non-thermal mechanisms that in- where ∂ Ĥq /∂λ is the qubit transverse susceptibility to
crease the generation rates, “bottleneck effects” that oc- fluctuations δλ, such that |δλ|2 is the ensemble average
cur at millikelvin temperatures to reduce recombination value of the environmental noise sources as seen by the
rates, or a combination of both. qubit. Eq. (54) is equivalent to Fermi’s Golden Rule,
It has been shown that the observed T1 and excess in which the qubit’s transverse susceptibility to noise is
excited-state population measured in today’s state-of- driven by the noise power spectral density. The qubit
the-art high-coherence transmon are self-consistent with transverse susceptibility can be used to calculate the pref-
excess “hot” nonequilibrium quasiparticles at the quasi- actors; for example, for fluctuations δλ = δn, the rele-
universal density of around 10−7 − 10−6 per Cooper vant term in the transmon Hamiltonian in Eq. (16) is
 20

(a) CP / CPMG cally modulate the transition frequency of the qubit and
2 2
thereby introduce a stochastic phase evolution of a qubit
2N N N N N N 2N
superposition state. This gradually leads to a loss of
t
phase information, and it is therefore called pure dephas-
N -pulses (N >1) ing (time constant Tϕ ). Unlike T1 relaxation, which is
1
generally an irreversible (incoherent) error, pure dephas-
s ing Tϕ is in principle reversible (a coherent error). The
N=0 CPMG degree of pure dephasing depends on the control pulse
sequence applied while the qubit is subject to the noise
)

1 process.
2 Consider the relative phase ϕ of a superposition state
6 10
g (
N

undergoing free evolution in the presence of noise. The

S~1/f superposition state’s accumulated phase,
Z t
ϕ(t) = ωq dt0 = hωq it + δϕ(t) (55)
0 0
0 2 4 6 8
Frequency, f (MHz)
(b) diffuses due to adiabatic fluctuations of the transition
frequency,
2T 1
∂ωq t
Z
10 δϕ(t) = δλ(t0 )dt0 , (56)
CPMG simulation ∂λ 0

where ∂ωq /∂λ = (1/~)|h∂ Ĥq /∂λi| is the qubit’s longi-

T2 ( s)

CPMG tudinal sensitivity to λ-noise. For noise generated by a

large number of fluctuators that are weakly coupled to
1
the qubit, its statistics are Gaussian. Ensemble averaging
over all realizations of the Gaussian-distributed stochas-
Ramsey (N=0) tic process δλ(t), the dephasing is
1 2

0.1 hei δϕ(t) i = e− 2 hδϕ (t)i

≡ e−χN (t) , (57)
1 10 100 1000
Number of pulses, N leading to a coherence decay function,

FIG. 7. Dynamical error suppression. (a) Carr-Purcell-

τ ∂ωq ∞
 2 Z 
Meiboom-Gill (CPMG) pulse sequence applies N equally he−χN (τ )
i = exp − gN (ω, τ )S(ω)dω ,
spaced π pulses within an otherwise free-evolution time τ . 2 ∂λ −∞
Pulses in the time domain correspond to bandpass filters in (58)
the frequency domain (lower panel) which serve to shape the where g(ω, τ ) is a dimensionless weighting function.
noise power spectrum seen by the qubit. The centroid of the The function gN (ω, τ ) can be viewed as a frequency-
bandpass filter shifts to higher frequencies as N is increased. domain filter of the noise Sλ (ω) [see Fig. 7(a)]. In gen-
For noise that decreases with frequency, such as 1/f noise, eral, its filter properties depend on the number N and
larger N corresponds to less integrated noise impinging on
distribution of applied pulses. For example, considering
the qubit. (b) CPMG pulse sequence applied to a flux qubit
biased at a point that is highly sensitive to 1/f flux noise. The sequences of π-pulses78,154–158 ,
Ramsey (N = 0) time is approximately 300 ns, and the Hahn
echo (N = 1) time is approximately 1.5 µs. Increasing the 1 
gN (ω, τ ) = 1 + (−1)1+N exp(iωτ )+
number of CPMG pulses continues to increase the effective (ωτ )2
T2 time towards the 2T1 limit. Adapted from Ref. 78. N 2
X
2 (−1)j exp(iωδj τ ) cos(ωτπ /2) , (59)

j=1
4EC (n̂ − ng )2 , where we allow for an offset charge ng ,
and the susceptibility is given by 8EC n̂. We refer the where δj ∈ [0, 1] is the normalized position of the centre
reader to Refs. 151–153 for more details. of the jth π-pulse between the two π/2-pulses, τ is the
total free-induction time, and τπ is the length of each
π-pulse157,158 , yielding a total sequence length τ + N τπ .
2. Connecting Tϕ to S(ω) As the number of pulses increases for fixed τ , the filter
function’s peak shifts to higher frequencies, leading to a
If the coupling to the qubit is instead longitudinal, reduction in the net integrated noise for 1/f α -type noise
e.g. Ĥq is of the type σz or a† a, the noise will stochasti- spectra with α > 0. Similarly, for a fixed N , the filter
 21

function will shift in frequency with τ . Additionally, for a during free evolution – periods of time for which no con-
fixed time separation τ 0 = τ /N (valid for N ≥ 1), the fil- trol is applied to the qubit, except for very short dynami-
ter sharpens and asymptotically peaks at ω 0 /2π = 1/2τ 0 cal decoupling pulses – and during driven evolution – pe-
as more pulses are added. gN (ω, τ ) is thus called the “fil- riods of time during which the control fields are applied
ter function”78,156 , and it depends on the pulse sequences to the qubit. Both free-evolution and driven-evolution
being applied. From Eq. (58), the pure dephasing de- noise is important to characterize, as the noise PSD may
cay arises from a noise spectral density that is “shaped” differ for these two types of evolution, and both are uti-
or “filtered” by the sequence-specific filter function. By lized in the context of universal quantum computation.
choosing the number of pulses, their rotation axes, and We refer the reader to Ref. 132 for a summary of noise
their arrangement in time, we can design filter functions spectroscopy during both types of evolution.
that minimize the net noise power for a given noise spec- The Ramsey frequency itself is sensitive to longitudinal
tral density within the experimental constraints of the ex- noise, and monitoring its fluctuations is one means to
periment (e.g., pulse-modulation bandwidth of the elec- map out the noise spectral density over the sub-millihertz
tronics used to control the qubits). to ∼ 100 Hz range131,163 .
To give a standard example, we compare the coherence
integral for two cases: a Ramsey pulse sequence and a At higher frequencies, the CPMG dynamical decou-
Hahn echo pulse sequence. Both sequences involve two pling sequence can be used to create narrow-band fil-
π/2 pulses separated by a time τ , during which free evo- ters that “sample” the noise at different frequencies as a
lution of the qubit occurs in the presence of low-frequency function of the free-evolution time τ and the number of
dephasing noise. The distinction is that the Hahn echo pulses N . This has been used to map out the noise PSD
will place a single π pulse (N = 1) in the middle of the in the range 0.1 - 300 MHz78 . One must be careful of
free-evolution period, whereas the Ramsey does not use the additional small peaks at higher-frequencies, which
any additional pulses (N = 0). The resulting filter func- all contribute to the dephasing used to perform the noise
tions are: spectroscopy164 .
In fact, using pulse envelopes such as Slepians165 –
which are designed to have concentrated frequency re-
ωτ
g0 (ω, τ ) = sinc2 (60) sponse – to perform noise spectroscopy is one means to
2 reduce such errors157 .
2 ωτ ωτ
g1 (ω, τ ) = sin sinc2 (61) At even higher frequencies, measurements of T1 can be
4 4
used in conjunction with Fermi’s golden rule to map out
where the subscript N = 0 and N = 1 indicate the num- the transverse noise spectrum above 1 GHz62,78,166 .
ber of π-pulses applied for the Ramsey and Hahn echo The aforementioned are all examples of noise spec-
experiments, respectively. The filter function g0 (ω, τ ) for troscopy during free evolution. Noise spectroscopy dur-
the Ramsey case is a sinc-function centered at ω = 0. For ing driven evolution was also demonstrated using a “spin-
noise that decreases with frequency, e.g., 1/f flux noise locking” technique, where a strong drive along x or y
in superconducting qubits, the Ramsey experiment win- axes defines a new qubit quantization axis, whose Rabi
dows through the noise in S(ω) where it has its highest frequency is the new qubit frequency in the spin-locking
value. This is the worst choice of filter function for 1/f frame. The spin-locking frame is then used to infer the
noise. In contrast, the Hahn echo filter function has a noise spectrum while the qubit is continually subject to
centroid that is peaked at a higher frequency, away from a driving field. For more information, we refer the reader
ω = 0. In fact, it has zero value at ω = 0. For noise to Ref. 132.
that decreases with frequency, such as 1/f noise, this is
advantageous. This concept extends to larger numbers
N of π pulses, and is called a Carr-Purcell-Meiboom-Gill
(CPMG) sequence159,160 . In Fig. 7(b), the T2 time of a
qubit under the influence of strong dephasing noise is in- E. Engineering noise mitigation
creased toward the 2T1 limit using a CPMG dynamical
error-suppression pulse sequence with an increasing num- Here, we briefly review a few examples of techniques
ber of pulses, N . We refer the reader to Refs. 78, 161, and that have been developed to reduce noise or reduce its im-
162, where these experiments were performed with super- pact on decoherence (sensitivity). We stress that improv-
conducting qubits. ing gate fidelity is a comprehensive optimization task, one
that is full of trade-offs. It is thus important to identify
what the limiting factors are, what price we have to pay
3. Noise spectroscopy to diminish these limiting factors, and what advantage we
can achieve until reaching a better trade-off. These all
The qubit is highly sensitive to its noisy environment, require an accurate understanding the limitations on the
and this feature can be used to map out the noise power gate fidelity, the sources of decoherence, the properties
spectral density. In general, one can map the noise PSD of the noise, and how it affects the system performance.
 22

1. Materials and fabrication improvements Returning to excess quasiparticles, it has been shown
that quasiparticles can be stochastically pumped away
Numerous efforts have been undertaken to reduce from the qubit region, resulting in longer, and more sta-
noise-induced defects due to materials and fabrica- ble T1 times122 . Although the pumping technique uses
tion40,167 . In the case of charge noise, significant efforts a series of π-pulses, this technique differs from dynami-
have been made to reduce the number of defects, such cal error suppression of coherent errors in that pulses are
as substrate cleaning59,168 , substrate annealing169 , and stochastically applied, and that it addresses incoherent
trenching41,61 . In the case of flux noise, several groups errors (T1 ).
have performed experiments to characterize the behavior
and properties of magnetic-flux defects137,170,171 . More
recently, a number of groups have tried optical surface 4. Cryogenic engineering
treatments to remove these defects145 .
In the context of residual quasiparticles, it has been In the case of photon shot-noise, in addition to ap-
shown that adding quasiparticle traps to the circuit de- plying dynamical decoupling techniques, there have been
sign can reduce the quasiparticle number, particularly in several recent works aimed at reducing the thermal pho-
devices that create excess quasiparticles, such as classi- ton flux that reaches the device. This include optimizing
cal digital logic or operation in the presence of thermal the attenuation of the cryogenic setup107,149,173 , remak-
radiation172 ing the cryogenic attenuators with more efficient heat
sinking148 , adding absorptive “black” material to absorb
stray thermal photons174,175 , and adding additional cav-
2. Design improvements ity filters for thermalization176 .

Another strategy is to reduce qubit sensitivity to the

noise by design. A qubit can only lose energy to defects IV. QUBIT CONTROL
if it couples to them. It has been demonstrated that
altering the capacitor geometry to increase the electric- In this section, we will introduce how superconduct-
field mode volume reduces the electric field density in the ing qubits are manipulated to implement quantum al-
thin dielectric regions that cause loss. This effectively gorithms. Since the transmon-like variety of supercon-
reduces the “participation” of the defects and makes the ducting qubits has so far been the most widely deployed
qubits less senstivie to these noise sources.55,62,134 . modality for implementing quantum programs, the dis-
In another example, the split transmons built using cussion throughout this section will be focused on modern
asymmetric junctions have lower sensitivity to flux noise techniques for transmons. Nonetheless, the techniques
than their symmetric counterparts at the expense of de- introduced here are applicable to all types of supercon-
creased frequency tunability69 . This is a good trade-off ducting qubits.
to make, because generally one is interested in tuning the We start with a brief review of the gates used in clas-
qubit frequency over a somewhat restricted range (typ- sical computing as well as quantum computing, and the
ically around 1 GHz) about the qubit frequency. When concept of universality. Subsequently we discuss the most
such asymmetric transmons are used in a gate scheme common technique of driving single qubit gates via a
such as the adiabatic CPHASE-gate65 , (see Sec.IV F) the capacitive coupling of a microwave line, coupled to the
qubit is less sensitive to flux noise, has a lower dephasing qubit. We introduce the notion of “virtual Z gates” and
rate, and this should improve the gate fidelity in general. “DRAG” pulsing. In the latter part of this section, we re-
view some of the most common implementations of two-
qubit gates in both tunable and fixed-frequeny transmon
3. Dynamical error suppression qubits. The single-qubit and two-qubit operations to-
gether form the basis of many of the medium-scale su-
As introduced in the previous section, it is advan- perconducting quantum processors that exist today.
tageous to leverage the 1/ω distribution of flux noise, Throughout this section, we write everything in the
wherein a considerable amount of the noise power re- computational basis {|0i, |1i} where |0i is the +1 eigen-
sides at low frequencies, and so the noise is “quasi-static”. state of σz and |1i is the −1 eigenstate. We use capi-
The spin-echo technique116 , which disrupts the free evo- talized serif-fonts to indicate the rotation operator of a
lution by a π-pulse, is extremely effective in mitigating qubit state, e.g. rotations around the x-axis by an angle
the pure dephasing by refocusing the coherent phase dis- θ is written as
persion due to low-frequency noise. The more advanced θ
versions, such as the CPMG-sequence, use multiple π- Xθ = RX (θ) = e−i 2 σx = cos(θ/2)1 − i sin(θ/2)σx (62)
pulses to interrupt the system more frequently, pushing
the filter band to even higher frequencies – a technique and we use the shorthand notation ‘X’ for a full π rotation
known as dynamical decoupling 78 . about the x axis (and similarly for Y := Yπ and Z := Zπ ).
 23

CIRCUIT A. Boolean logic gates used in classical computers

GATE SYMBOL TRUTH TABLE

Universal boolean logic can be implemented on classi-

NOT The output is 1 Input Output cal computers using a small set of single-bit and two-bit
when the input
is 0 and 0 when 0 1 gates. Several common classical logic gates are shown in
the input is 1. 1 0 Fig. 8 along with their truth tables. In classical boolean
logic, bits can take on one of two values: state 0 or state
1. The state 0 represents logical FALSE, and state 1 rep-
resents logical TRUE.
Input Output
AND The output is 1
Beyond the trivial “identity operation,” which simply
0 0 0 passes a boolean bit unchanged, the only other possible
only when both
inputs are 1, 0 1 0 single-bit boolean logic gate is the NOT gate. As shown
otherwise the 1 0 0 in Fig. 8, the NOT gate flips the bit: 0 → 1 and 1 → 0.
output is 0. 1 1 1 This gate is reversible, because it is trivial to determine
the input bit value given the output bit values. As we
Input Output will see, for two-bit gates, this is not the case.
OR The output is 0 0 0 0 There are several two-bit gates shown in Fig. 8. A
only when both
inputs are 0, 0 1 1 two-bit gate takes two bits as inputs, and it passes as
otherwise the 1 0 1 an output the result of a boolean operation. One com-
output is 1. 1 1 1 mon example is the AND gate, for which the output is
1 if and only if both inputs are 1; otherwise, the output
Input Output is 0. The AND gate, and the other two-bit gates shown
NAND The output is 0 0 0 1 in Fig. 8, are all examples of irreversible gates; that is,
only when both
0 1 1 the input bit values cannot be inferred from the output
inputs are 1, values. For example, for the AND gate, an output of log-
otherwise the 1 0 1
output is 1. 1 1 0 ical 1 uniquely identifies the input 11, but an output of 0
could be associated with 00, 01, or 10. Once the opera-
tion is performed, in general, it cannot be “undone” and
Input Output
the input information is lost. There are several variants
NOR The output is 1
only when both
0 0 1 of two-bit gates, including,
0 1 0
inputs are 0,
otherwise the 1 0 0 • AND and OR;
output is 0. 1 1 0

• NAND (a combination of NOT and AND) and NOR

Input Output
XOR The output is 1 (a combination of NOT and OR);
only when the two 0 0 0
inputs have 0 1 1
different value, • XOR (exclusive OR) and NXOR (NOT XOR).
1 0 1
otherwise the
1 1 0
output is 0. The XOR gate is interesting, because it is a parity gate.
That is, it returns a logical 0 if the two inputs are the
Input Output
XNOR The output is 1 same values (i.e., they have the same parity), and it re-
only when the two 0 0 1 turns a logical 1 if the two inputs have different values
inputs have the 0 1 0 (i.e., different parity). Still, the XOR and NXOR gates
same value,
otherwise the
1 0 0 are not reversible, because knowledge of the output does
output is 0. 1 1 1 not allow one to uniquely identify the input bit values.
The concept of universality refers to the ability to per-
FIG. 8. Classical single-bit and two-bit boolean logic gates.
form any boolean logic algorithm using a small set of
For each gate, the name, a short description, circuit repre- single-bit and two-bit gates. A universal gate set can in
sentation, and input/output truth tables are presented. The principle transform any state to any other state in the
numerical values in the truth table correspond to the classical state space represented by the classical bits. The set of
bit values 0 and 1. Adapted from Ref. 177. gates which enable universal computation is not unique,
and may be represented by a small set of gates. For ex-
ample, the NOT gate and the AND gate together form
a universal gate set. Similarly, the NAND gate itself is
universal, as is the NOR gate. The efficiency with which
one can implement arbitrary boolean logic, of course, de-
pends on the choice of the gate set.
 24

CIRCUIT MATRIX TRUTH BLOCH

GATE REPRESENTATION REPRESENTATION TABLE SPHERE

z
I Identity-gate: Input Output
no rotation is
performed.
I I= ( ) 1 0
0 1 |0〉 |0〉
|1〉 |1〉
y
x

z
X gate: Input Output
rotates the
qubit state by
X X= ( ) 0 1
1 0 |0〉 |1〉
π radians |1〉 |0〉
(180º) about 180º
the x-axis. y
x

z
Y gate: Input Output
rotates the
qubit state by Y Y= ( ) 0 –i
i 0 |0〉 i |1〉
π radians 180º
(180º) about |1〉 –i |0〉
the y-axis. y
x

180º
z
Z gate: Input Output
rotates the
qubit state by
Z Z= ( ) 1 0
0 –1 |0〉 |0〉
π radians |1〉 –|1〉
(180º) about
the z-axis. y
x

90º z
S gate: Input Output
rotates the
qubit state by π
2
S S= (1 0π
0 ei 2 ) |0〉 |0〉
radians (90º) π
about the
|1〉 e i 2|1〉
z-axis. y
x

T gate: 45º z
Input Output
rotates the
qubit state by π
4
T T= ( 1 0π
0 ei 4 ) |0〉 |0〉
π
radians (45º) |1〉 e i 4|1〉
about the
z-axis. y
x

H gate: Input Output

z
rotates the
qubit state by
π radians (180º)
H H=
1
2
( )
1 1
1 –1
|0〉 |0〉 + |1〉
180º
2
about an axis
diagonal in the |1〉 |0〉 – |1〉
x-z plane. This is 2 y
equivalent to an
X-gate followed x
by a π2 rotation
about the y-axis.

FIG. 9. Quantum single-qubit gates. For each gate, the name, a short description, circuit representation, matrix representation,
input/output truth tables, and Bloch sphere represenation are presented. Matrices are defined in the basis spanned by the
state vectors |0i ≡ [1 0]T and |1i ≡ [0 1]T . The numerical values in the truth table correspond to the quantum states |0i and
|1i. Adapted from Ref. 177.

B. Quantum logic gates used in quantum computers course assume the classical states |0i and |1i, at the north

Quantum logic can similarly be performed by a small

set of single-qubit and two-qubit gates. Qubits can of
 25

CIRCUIT MATRIX TRUTH

GATE REPRESENTATION REPRESENTATION TABLE

Input Output
Controlled-NOT gate: 1 0 0 0
apply an X-gate to the |00〉 |00〉
0 1 0 0
target qubit if the CNOT = |01〉 |01〉
0 0 0 1
control qubit is in state 0 0 1 0 |10〉 |11〉
|1〉 |11〉 |10〉

Input Output
Controlled-phase gate: 1 0 0 0
apply a Z-gate to the |00〉 |00〉
0 1 0 0
target qubit if the CPHASE = |01〉 |01〉
0 0 1 0
control qubit is in |10〉 |10〉
state |1〉 Z 0 0 0 –1
|11〉 |11〉

FIG. 10. Quantum two-qubit gates: the controlled NOT (CNOT) gate and the controlled phase (CPHASE or CZ). For each
gate, the name, a short description, circuit representation, matrix representation, and input/output truth tables are presented.
Matrices are defined in the basis spanned by the two-qubit state vectors |00i ≡ [1 0 0 0]T , |01i ≡ [0 1 0 0]T , |10i ≡ [0 0 1 0]T ,
and |11i ≡ [0 0 0 1]T , where the first qubit is the control qubit, and the second qubit is the target qubit. The CNOT gate flips
the state of the target qubit conditioned on the control qubit being in state |1i. The CPHASE gate applies a Z gate to the
target qubit conditioned on the control qubit being in state |1i. Adapted from Ref. 177.

pole and south pole of the Bloch sphere, but they can also can be built from the CNOT-gate and single-qubit gates.
assume arbitrary superpositions α|0i + β|1i, correspond-
ing to any other position on the sphere.
Single-qubit operations translate an arbitrary quan-
tum state from one point on the Bloch sphere to another
point by rotating the Bloch vector (spin) a certain an-
gle about a particular axis. As shown in Fig. 9, there
are several single-qubit operations, each represented by
a matrix that describes the quantum operation in the
computational basis represented by the eigenvectors of
the σz operator, i.e. |0i ≡ [1 0]T and |1i ≡ [0 1]T .
For example, the identity gate performs no rotation on
the state of the qubit. This is represented by a two-by-
two identity matrix. The X-gate performs a π rotation
about the x axis. Similarly, the Y-gate and Z-gate per-
form a π rotation about the y axis and z axis, respec-
tively. The S-gate performs a π/2 rotation about the z
axis, and the T-gate performs a rotation of π/4 about
the z axis. The Hadamard gate H is also a common
single-qubit gate the performs a π rotation about an axis
diagonal in the x-z plane, see Fig. 9.
Two-qubit quantum-logic gates are generally condi-
tional gates and take two qubits as inputs. Typically,
the first qubit is the control qubit, and the second is the
target qubit. A unitary operator is applied to the target
qubit, dependent on the state of the control qubit. The
two common examples shown in Fig. 10 are the controlled
NOT (CNOT-gate) and controlled phase (CZ or CPHASE
gate). The CNOT-gate flips the state of the target qubit
conditioned on the control qubit being in state |1i. The
CPHASE-gate applies a Z gate to the target qubit, condi-
tioned on the control qubit being in state |1i. As we will
shown later, the iSWAP gate – another two-qubit gate –
 26

The unitary operator of the CNOT gate can be written (a) (b)
Classical NOT gate Quantum X gate
in a useful way, highlighting that it applies an X depend-
ing on the state of the control qubit. b b

1 0 0 0 in out
 
0 1 0 0
UCNOT = = |0ih0| ⊗ 1 + |1ih1| ⊗ X (63)
0 0 0 1 b b
0 0 1 0

and similarly for the CPHASE gate, FIG. 11. Comparison of the classical inverter (NOT) gate
and quantum bit flip (X) gate. (a) The classical NOT gate
1 0 0 0
 
that inverts the state of a classical bit. (b) The quantum X
0 1 0 0 gate, which flips the amplitudes of the two components of a
UCPHASE = = |0ih0| ⊗ 1 + |1ih1| ⊗ Z (64)
0 0 1 0 quantum bit.
0 0 0 −1

Comparing the last equality above with the unitary for A universal set of single-qubit and two-qubit gates is
the CNOT [Eq. (63)], it is clear that the two gates are sufficient to implement arbitrary quantum logic. This
closely related. Indeed, a CNOT can be generated from means that this gate set can in principle reach any state
a CPHASE by applying two Hadamard gates, in the multi-qubit state-space. How efficiently this is
done depends on the choice of quantum gates that com-
UCNOT = (1 ⊗ H)UCPHASE (1 ⊗ H), (65) prise the gate set. We also note that each of the single-
qubit and two-qubit gates is reversible, that is, given the
since HZH = X. Due to the form of Eq. (64), the CPHASE output state, one can uniquely determine the input state.
gate is also denoted the CZ gate, since it applies a con- As we discuss further, this distinction between classical
trolled Z operator, by analogy with CNOT (a controlled and quantum gates arises, because quantum gates are
application of X operator). Inspection of the definition of based on unitary operations U . If a unitary operation U
CPHASE in Fig. 10 makes no distinction between which is a particular gate applied to a qubit, then its hermitian
qubit acts as the target and which as the control and, conjugate U † can be applied to recover the original state,
consequently, the circuit-diagram is sometimes drawn in since U † U = I resolves an identity operation.
a symmetric fashion
•
CPHASE = • (66)
C. Comparing classical and quantum gates
The CNOT in terms of CPHASE can then be realized as
The gate-sequences used to represent quantum algo-
• rithms have certain similarities to those used in classi-
CNOT = (67) cal computing, with a few striking differences. As an
H • H example, we consider first the classical NOT gate (dis-
cussed previously), and the related quantum circuit ver-
Some two-qubit gates such as CNOT and CPHASE are
sion, shown in Fig. 11.
also called entangling gates, because they can take prod-
While the classic bit-flip gate inverts any input state,
uct states as inputs and output entangled states. They
the quantum bit-flip does not in general produce the an-
are thus an indispensable component of a universal gate
tipodal state (when viewed on the Bloch sphere), but
set for quantum logic. For example, consider two qubits
rather exchange the prefactors of the wavefunction writ-
A and B in the following state:
ten in the computational basis. The X operator is some-
1 times referred to as ‘the quantum NOT’ (or ‘quantum
|ψi = √ (|0i + |1i)A |0iB . (68) bit-flip‘), but we note that X only acts similar to the
2
classical NOT gate in the case of classical data stored in
If we perform a CNOT gate, UCNOT , on this state, with the quantum bit, i.e. X|gi = |ḡi for g ∈ {0, 1}.
qubit A the control qubit, and qubit B the target qubit, As briefly mentioned in Sec. IV B, all quantum gates
the resulting state is (see the truth table in Fig. 10): are reversible, due to the underlying unitary nature of
the operators implementing the logical operations. Cer-
1 tain other processes used in quantum information pro-
UCNOT |ψi = √ (|0iA |0iB + |1iA |1iB ) 6= (. . .)A (. . .)B ,
2 cessing, however, are irreversible. Namely, measurements
(69) (see Sec. V for detailed discussion) and energy loss to the
which is a state that cannot be factored into an isolated the environment (if the resulting state of the environment
qubit-A component and a qubit-B component. This is is not known). Here, we will not consider how these pro-
one of the two-qubit entangled Bell states, a manifestly cesses are modeled, but refer the interested reader to e.g.
quantum mechanical state. Ref. 178, and will only consider unitary control oper-
 27

ations throughout the rest of this section. Finally, we as possible, and one wants to use as many of the native
note that quantum circuits are written left-to-right (in gates as possible, to reduce the amount of time spent syn-
order of application), while the calculation of the result thesising. Moreover, running a quantum algorithm also
of a gate-sequences, e.g the circuit depends on the qubit connectivity of the device. The
process of designing a quantum gate sequence that effi-
|ψin i U0 U1 ··· Un |ψout i (70) ciently implements a specific algorithm, while taking into
account the considerations outlined above is known as
is performed right-to-left, i.e. gate synthesis and gate compilation, respectively. A full
discussion of this large research effort is outside the scope
|ψout i = Un · · · U1 U0 |ψin i. (71) of this review, but the interested reader may consult e.g.
Refs. 183–185 and references therein as a starting point.
As discussed in Sec. IV A, the NOR and NAND gates are As a concrete (and trivial) example of how gate identities
each individually universal gates for classical computing. can be used, in Eq. (74) we illustrate how the Hadamard
Since both of these gates have no direct quantum ana- gate from G1 can be generated by two single-qubit gates
logue (because they are not reversible), it is natural to (from G0 ) and an overall phase gate,
ask which gates are needed to build a universal quan-
tum computer. It turns out that the ability to rotate
about arbitrary axes on the Bloch-sphere (i.e. a complete 1 1 −1 −i 0 1 1 1
    
H = Ph Y Zπ = i √
π π =√
single-qubit gate set), supplemented with any entangling 2 2
2 1 1 0 i 2 1 −1
2-qubit operation will suffice for universality178,179 . By (74)
using what is known as the ‘Krauss-Cirac decomposition’,
any two-qubit gate can be decomposed into a series of As we show in Sec. IV D 1, the gates Xθ , Yθ and Zθ
CNOT operations178,180 . are all natively available in a superconducting quantum
processor.
We now address the question of how single qubit ro-
1. Gate sets and gate synthesis tations and two-qubit operations are implemented in
transmon-based superconducting quantum processors.
A common universal quantum gate set is

G0 = {Xθ , Yθ , Zθ , Phθ , CNOT} (72) 2. Addressing superconducting qubits

where Phθ = eiθ 1 applies an overall phase θ to a single

The modes of addressing transmon-like superconduct-
qubit. For completeness we mention another universal
ing qubits can roughly be split into two main categories:
gate set which is of particular interest from a theoretical
i) Capacitive coupling between a resonator (or a feedline)
perspective, namely
and the superconducting qubit dipole-field allows for mi-
G1 = {H, S, T, CNOT}, (73) crowave control to implement single-qubit rotations (see
Sec. IV D) as well as certain two-qubit gates (see Sec-
As a technical aside, we mention that the restriction to tions IV G and IV G 4). ii) For flux-tunable qubits, local
a discrete gate set still gives rise to universality. This magnetic fields can be used to tune the frequency of indi-
fact relies on using the so-called Solovay-Kitaev181,182 vidual qubits. This allows the implementation of z-axis
theorem, which (roughly) states that any other single- single-qubit rotation as well as multiple two-qubit gates
qubit gate can be approximated to an error  using only (see Sections IV E, IV F and IV H).
O(logc (1/)) (where c > 0) single-qubit gates from G1 .
The gate-set G1 is typically referred to as the ‘Clifford +
T ’ set, where H, S and CNOT are all Clifford gates. D. Single-qubit gates
Each quantum computing architecture will have cer-
tain gates that are simpler to implement at the hard- In this section we will review the steps necessary to
ware level than others (sometimes referred to as ’na- demonstrate that capacitive coupling of microwaves to a
tive’ gates of the architecture). These are typically the superconducting circuit can be used to drive single-qubit
gates for which the Hamiltonian governing the gate- gates. To this end we consider coupling a superconduct-
implementation gives rise to a unitary propagator that ing qubit to a microwave source (sometimes referred to
corresponds to the gate itself. We will show several exam- as a ‘qubit drive’) as shown in Fig. 12(a). A full circuit
ples of this in Sections IV E, IV F, and IV G. Regardless analysis of the circuit in Fig. 12(a) is beyond the scope
of which gates are natively available, as long as one has a of this review, so here we settle for highlighting the steps
complete gate set, one can use the Solovay-Kitaev theo- that elucidate the physics of the qubit/drive coupling.
rem to synthesize any other set efficiently. In general one The interested reader may consult a number of lectures
wants to keep the overall number of time steps in which notes and pertinent theses (e.g. Refs. 44, 163, 186–188).
gates are applied (denoted the depth of a circuit) as low Here we follow Ref. 163.
 28

Vd (t) Rw Cd
To elucidate the role of the drive, we move into a frame
rotating with the qubit at frequency ωq (also denoted
‘the rotating frame’ or the ‘the interaction frame’). To
see the usefulness√ of this rotating frame, consider a state
C |ψ0 i = (1 1)T / 2. By the time-dependent Schrödinger
equation this state evolves according to
room 1
 iω t/2 
e q
temperature wiring on-chip |ψ0 (t)i = UH0 |ψ0 i = √ −iω t/2 , (80)
2 e q
FIG. 12. Circuit diagram of capacitive coupling of a mi- where UH0 is the propagator corresponding to H0 . By
crowave drive line (characterized by a time-dependent voltage calculating e.g. hψ0 |σx |ψ0 i = cos(ωq t) it is evident
Vd (t)) to a generic transmon-like superconducting qubit. that the phase is winding with a frequency of ωq due
to the σz term. By going into a frame rotating with
the qubit at frequency ωq , the action of the drive can
1. Capacitive coupling for X,Y control
be more clearly appreciated. To this end we define
†
Urf = eiH0 t = UH and the new state in the rotating
We start by modeling the qubit as an harmonic os- 0
frame is |ψrf (t)i = Urf |ψ0 i. The time-evolution in this
cillator, for which the (classical) circuit Hamiltonian can new frame is again found from the Schrödinger equation
be calculated by circuit quantization techniques, starting (using the shorthand ∂t = ∂/∂t),
from Kirchoffs laws, and is given by163
i∂t |ψrf (t)i = i(∂t Urf )|ψ0 i + iUrf (∂t |ψ0 i) (81)
Q̃(t)2 Φ2 Cd
H= + + Vd (t)Q̃, (75) = iU̇rf Urf† |ψrf i
+ Urf H0 |ψ0 i (82)
2CΣ 2L CΣ
 
where CΣ = C + Cd is the total capacitance to ground = iU̇rf Urf† + Urf H0 Urf† |ψrf i (83)
and Q̃ = CΣ Φ̇ − Cd Vd (t) is a renormalized charge vari- | {z }
able for the circuit. We can now promote the flux and He0
charge variables to quantum operators and assume weak
coupling to the drive-line, so that Q̃ ≈ Q̂, and arrive at We can think of the term H e 0 in the parentheses in
Cd Eq. (83) as the form of H0 in the rotating frame. Sim-
H = HLC + Vd (t)Q̂, (76) ple insertion shows that H e 0 = 0 as expected (the ro-
CΣ
tating frame should take care of the time-dependence).
where HLC = Q̂2 /(2C) + Φ̂2 /(2L) and we have kept However, one could also think of the term in brackets
only terms that couple to the dynamic variables. Similar in Eq. (83) as a prescription for calculating the form of
to the momentum operator for a harmonic oscillator in any Hamiltonian in the rotating frame given by Urf , by
(x, p)–space, we can express the charge variable in terms replacing H0 with some other H. In general, we will not
of raising and lowering operators, as done in Sec. II find H
e = 0.
Q̂ = −iQzpf a − a†


(77) Returning to Eq. (79), the form of Hd in the rotating
frame is found to be
where Qzpfp = ~/2Z is the zero-point charge fluctations
p
e d = ΩVd (t) (cos(ωq t)σy − sin(ωq t)σx ) .
H (84)
and Z = L/C is the impedance of the circuit to ground.
Thus, the LC oscillator capacitively coupled to a drive We can in general assume that the time-dependent part
line can be written as, of the voltage (Vd (t) = V0 v(t)) has the generic form
1
 
Cd
H =ω a a+†
Vd (t)iQzpf a − a† . (78) v(t) = s(t) sin(ωd t + φ) (85)


−
2 CΣ
= s(t) (cos(φ) sin(ωd t) + sin(φ) cos(ωd t)) , (86)
Finally, by truncating to the lowest transition of the oscil-
lator we can make the replacement a → σ − and a† → σ + where s(t) is a dimensionless envelope function, so that
throughout and arrive at the amplitude of the drive is set by V0 s(t). Adopting the
ωq definitions
H = − σz + ΩVd (t)σy (79)
2 } | {z }
| {z I = cos(φ) (the ‘in-phase’ component) (87)
H0 Hd
Q = sin(φ) (the ‘out-of-phase’ component) (88)
where Ω = (Cd /CΣ )Qzpf and ωq = (E1 − E0 )/~.§

§ Starting from a generic qubit Hamiltonian, H0 = E0 |0ih0| + H0 = ((E0 + E1 )/2)1 − ((E1 − E0 )/2)σz . In the main text we
E1 |1ih1|, we can rewrite in terms of Pauli matrices, and get have ignored the constant offset term.
 29

the driving Hamiltonian in the rotating frame takes the (a) (b)
form
LO AWG
I
e d = ΩV0 s(t) (I sin(ωd t) − Q cos(ωd t)) carrier I Q baseband

Amplitude
H pulses
Q

× (cos(ωq t)σy − sin(ωq t)σx ) (89)

Performing the multiplication and dropping fast rotating ωLO s(t)

LO ωAWG
terms that will average to zero (i.e. terms with ωq + ωd ), time
I Q (c)
known as the rotating wave approximation (RWA), we RF
Vd(t)
are left with
-x(-I)
s(t)
-y(-Q) y(Q)
e d = 1 ΩV0 s(t) (−I cos(δωt) + Q sin(δωt)) σx

H x(I)
2 ωd=ωLO+ωAWG

+ (I sin(δωt) − Q cos(δωt)) σy (90) to qubits

FIG. 13. (a) Schematic of a typical qubit drive setup. A mi-

where δω = ωq − ωd . Finally, by re-using the definitions crowave source supplies a high-frequency signal (ωLO ), while
from Eq. (86), the driving Hamiltonian in the rotating an arbitrary waveform generator (AWG) supplies a pulse-
frame using the RWA can be written as envelope (s(t)), sometimes with a low frequency component,
ωAWG , generated by the AWG. The IQ-mixer combines the two
e d = − Ω V0 s(t) 0
 
ei(δωt+φ) signals to generate a shaped waveform Vd (t) with a frequency
H . (91)
2 e−i(δωt+φ) 0 ωd = ωLO ± ωAWG , typically resonant with the qubit. (b) Ex-
ample of how a gate sequence is translated into a waveform
Equation (91) is a powerful tool for understanding single- generated by the AWG. Colors indicate I and Q components.
qubit gates in superconducting qubits. As a concrete (c) The action of a Xπ/2 pulse on a |0i state to produce the
example, assume that we apply a pulse at the qubit fre- |−ii = √12 (|0i − i|1i) state.
quency, so that δω = 0, then

e d = − Ω V0 s(t) (Iσx + Qσy ) ,

H (92)
2 sequence of pulses (see Fig. 13(a)) Θk , Θk−1 , ...Θ0 is con-
verted to a sequence of gates operating on a qubit as
showing that an in-phase pulse (φ = 0, i.e. the I-
component) corresponds to rotations around the x-axis, k
e[− 2 Θn (t)(In σx +Qn σy )] ,
i
while an out-of-phase pulse (φ = π/2, i.e. the Q-
Y
Uk · · · U1 U0 = T (95)
component), corresponds to rotations about the y-axis. n=0
As a concrete example of an in-phase pulse, writing out
the unitary operator yields where T is an operator that ensures the pulses are gen-
erated in the time-ordered sequence corresponding to
Uk · · · U1 U0 .
 Z t  
φ=0 i
Urf,d (t) = exp ΩV0 s(t0 )dt0 σx , (93) In Fig. 13 we outline the typical IQ modulation setup
2 0
used to generate the pulses used in Eq. (95). Fig. 13(a)
which depends only on the macroscopic design parame- shows how a pulse at frequency ωd is generated using a
ters of the circuit as well as the envelope of the baseband low phase-noise microwave generator (typically denoted
pulse s(t) and amplitude V0 , which can both be controlled ‘the local oscillator (LO)’), while the pulse is shaped by
using arbitrary waveform generators (AWGs). Equation combining the LO signal in an IQ mixer with pulses gen-
(93) is known as Rabi driving and can serve as a use- erated in an AWG. To allow for frequency multiplexing,
ful tool for engineering the circuit parameters needed for the AWG signal will typically be generated with a low-
efficient gate operation (subject to the available output frequency component, ωAWG , and the LO signal will be
voltage V0 ). To see this we define the shorthand offset, so that ωLO +ωAWG = ωd . By mixing in more than
one frequency ωAWG1 , ωAWG2 , ... it is possible to address
t
multiple qubits (or readout resonators) simultaneously,
Z
Θ(t) = −ΩV0 s(t0 )dt0 (94) via the superposition of individual drives.
0
The I (Q) input of the IQ mixer will multiply the base-
which is the angle by which a state is rotated given the band signal to the in-phase (out-of-phase) component of
capacitive couplings, the impedance of the circuit, the the LO. In Fig. 13(b) we schematically show the com-
magnitude V0 , and the waveform envelope, s(t). This parison between XY gates in a quantum circuit and the
means that to implement a π-pulse on the x-axis one corresponding waveforms generated in the AWG (omit-
would solve the equation Θ(t) = π and output the sig- ting for clarity the frequency ωAWG component). The
nal in-phase with the qubit drive. In this language, a inset in Fig. 13(b) shows an example of a gate on the
 30

Bloch sphere, with indication of (I, Q) axes. More so- Finally we mention one more salient feature of the
phisticated and compact approaches exist to reduce the virtual-Z gates. As shown in Ref.63, any single-qubit
hardware needed for XY qubit control, relative to the operation (up to a global phase) can be written as
setup shown in Fig. 13, see e.g.189–191 .
U (θ, φ, λ) = Zφ− π2 X π2 Zπ−θ X π2 Zλ− π2 , (100)

2. Virtual Z gate for appropriate choice of angles θ, φ, λ. This means

that access to a single physical X π2 combined with the
virtual-Z gives access to a complete single qubit gate
As we saw in Sec. IV D, the distinction between x–
set! An explicit example of Eq. (100) in action is the
and y–rotations was merely a choice of phase on the mi-
Hadamard gate, which can be written as H = Z π2 X π2 Z π2 ,
crowave signals, and the angle to be rotated is given by
but since the Z’s can be virtual, it is possible to imple-
Θ(t), both of which are generated using an AWG. Since
ment Hadamards as an effective single pulse operation in
the choice of phase φ has an arbitrary starting point, we
superconducting qubits.
could consider φ → φ + π/2. This would lead to I → Q
and Q → −I. Therefore, changing the phase effectively
changes rotations around x to rotations around y (and 3. The DRAG scheme
vice-versa, with a change of sign). This is reminiscent of
the result of applying a Zπ rotation to x– and y–rotations,
where Zπ Xπ = iYπ and Zπ Yπ = −iXπ . This analogy be- In going from Eq. (78) to Eq. (79) we assumed we
tween shifting a phase of an AWG-generated signal and could ignore the higher levels of the qubit. However,
applying Z rotations can be utilized to implement virtual for weakly anharmonic qubits, such as the transmon (see
Z gates192 . As shown by McKay et al., this intuition Sec. II), this may not be a justified assumption, since
can be formalized via the following example: consider ωq1→2 only differs from ωq (≡ ωq0→1 ) by the anharmonicity,
the case of applying a pulse with an angle θ on the I α = ωq1→2 − ωq , which is negative and typically around
channel (i.e. a Xθ ) followed by another θ pulse on the I 200 to 300 MHz. This situation is sketched in Fig. 14(a-
channel, but with a phase φ0 relative to the first pulse c), where we illustrate how Gaussian pulses with stan-
(φ )
(denoted Xθ 0 , where X indicates we still use the I chan- dard deviations σ = {1, 2, 5} ns have spectral content
nel, but the rotation axis is now an angle φ0 away from that leads to non-zero overlaps with the ωq1→2 = ωq − |α|
the x-axis). Using Eq. (95) this corresponds to a pulse frequency. This leads to two deleterious effects: (1 ) leak-
sequence age errors which take the qubit out of the computational
subspace, and (2 ) phase errors. Effect 1 can occur be-
Xθ
(φ0 ) θ
Xθ = e−i 2 (cos(φ0 )σx +sin(φ0 )σy ) Xθ (96) cause a qubit in the state |1i may be excited to |2i as a π
pulse is applied, or be excited directly from the |0i, since
= Z−φ0 Xθ Zφ0 Xθ (97)
the qubit spends some amount of time in the |1i state
from which we see that the effect of the offset phase φ0 during the π pulse. Effect 2 occurs because the presence
is to apply Zφ0 . The equality above can be verified with of the drive results in a repulsion between the |1i and |2i
a little trigonometric footwork. The final Z−φ0 is due to levels, in turn changing ωq0→1 as the pulse is applied. This
the rotation being in the frame of reference of the qubit. leads to the accumulation of a relative phase between
However, since readout is along z-axis (see Sec. V), a |0i and |1i194 . The so-called DRAG procedure195–197
final phase rotation about z will not change the mea- (Derivative Reduction by Adiabatic Gate) seeks to com-
surement outcome. Thus, if one wants to to implement bat these two effects by applying an extra signal in the
the gate sequence out-of-phase component. The trick is to modify the wave-
form envelope s(t) according to
··· Ui Zθ0 Ui+1 Zθ1 Ui+2 ··· (98) 
 s(t) on I
where Ui ’s are arbitrary gates, this can be done by re- s(t) → s0 (t) = , (101)
vising the gate sequence (in the control software for the  λ ṡ(t) on Q
α
AWG) and changing the phase of subsequent pulses
where λ is a dimensionless scaling parameter, and λ =
··· Ui
(θ )
Ui+10
(θ +θ1 )
Ui+20 ··· (99) 0 correspond to no DRAG pulse and ṡ(t) is the time
derivative of s(t). The theoretically optimal choice for
which reduces the number of overall gates. Moreover, reducing dephasing error is λ = 0.5 and an optimal choice
the virtual-Z gates are “perfect”, in the sense that no for reducing leakage error is λ = 1196,198 . Interchanging
additional pulses are required, and the gate takes “zero I and Q in Eq. (101) corresponds to DRAG pulsing for
time”, and thus the gate fidelity is nominally unity. As the Q component.
we show in Sections IV E and IV F, operation of two- In practice there can be a deviation from these two
qubit gates can incur additional single-qubit phases. Us- optimal values, often due to pulse distortions in the lines
ing the virtual-Z strategy, these phases can be cancelled leading to the qubits. Typically, randomized benchmark-
out, leaving a pure two-qubit interaction. ing experiments combined with single-shot measurements
 31

(a) (b) quency), i.e.

2E1 σ (ns)
1

Ampl. (a.u.)
α s0δf (t) = s0 (t)ei2πδf t , (102)
E2 2
5
and choosing λ to minimize leakage errors, then phase
errors can be reduced simultaneously199 . Similarly,
by a judicious use of the virtual-Z gate, it is also
E1 -20 -10 0 10 20 possible to reduce phase errors in combination with
(c) Time (ns) DRAG pulsing to reduce leakage192 . Modern single-qubit
FFT gates using DRAG pulsing now routinely reach fidelities

Ampl. (a.u.)
F1qb & 0.9965,67,199,202–205 . Other techniques also exist
for operating single-qubit gates in a spectrally crowded
device206,207 .
α
ωq01 E0

-0.5 -0.25 0 0.25 0.5 E. The iSWAP two-qubit gate in tunable qubits
Frequency (GHz)
(d) (e)
time As briefly mentioned in Sec. IV C, single-qubit gates
λ=0 I supplemented with an entangling two-qubit gate can
Ampl. (a.u.)

Q
form the gate set required for universal quantum compu-
tation. The two-qubit gates available in the transmon-
like superconducting qubit architecture can roughly be
split into two broad families as outlined previously:
one group requiring local magnetic fields to tune the
transition frequency of qubits and one group consist-
(f) time (ns) (g) ing of all-microwave control. There exist several hy-
λ = 0.5 I time brid schemes that combine various aspects of these two
Ampl. (a.u.)

Q categories and, in particular, the notions of tunable

coupling and parametric driving are proving to be im-
portant ingredients in modern superconducting qubit
processors63,67,91,104,107,208–214 . In this section, however,
we start by introducing the iSWAP gate, and then review
the CPHASE (controlled-phase) in Section IV F and the
time (ns) CR (cross-resonance) in Section IV G. We briefly review
a few other two-qubit gates and discuss their merits in
FIG. 14. (a) Schematic level diagram of a weakly anharmonic Sections IV G 4 and IV H.
transmon qubit subjected to a drive at transition frequency
ωd = ωq . (b) Gaussian waveform with standard deviation
σ. (c) Fourier transform of (b) showing how the short pulse 1. Deriving the iSWAP unitary
lengths lead to significant overlap with the ωq1→2 transition,
separated from ωq by the anharmonicity α. (d) Waveform
of a Xπ pulse without DRAG modulation. (e) Effect of the As we saw in Sec. II, Eq.32 the interaction term be-
waveform from (d) on a qubit initialized in the |0i state with tween two capacitively coupled qubits (in the two-level
α = −200 MHz and ωq = 4 GHz. The dephasing error is approximation) is given by
visible as a deviation from the |1i after the pulse. (f) Wave-
form of a Xπ pulse with DRAG modulation for a qubit with Hqq = gσy1 ⊗ σy2 , (103)
anharmonicity α = −200 MHz and DRAG parameter λ = 0.5 where g is the coupling strength and ⊗ is used to em-
to cancel dephasing errors (see text for details). (g) Effect of
phasize the tensor product. If the capacitive coupling is
the waveform from (f) on the same qubit as (e). Calculated
using mesolve in the software package QuTiP193 . mediated through a bus resonator, then215,216
g1 g2 (∆1 + ∆2 )
g → gq-r-q = , (104)
2∆1 ∆2
where gi is the resonator coupling to qubit i (depen-
(see Sec. V) of the |2i state is used to determine the op- dent on the qubit-resonator coupling capacitance Cqir )
timal value of λ. The λ = {0.5, 1} tradeoff was demon- and ∆i = ωqi − ωr is the detuning of qubit i to the res-
strated explicitly in192,199 . However, by extending the onator. In the simpler case where the qubits are directly
original DRAG pulse implementation200,201 , it is is possi- coupled217 ,
ble to reduce both errors simultaneously. By introducing
a frequency detuning parameter δf to the waveform196 1√ Cq-q
g → gq-q = ωq1 ωq2 p , (105)
(defined such that δf = 0 corresponds to qubit fre- 2 Cq-q + C1 Cq-q + C2
p
 32

where Cq-q is the qubit-qubit coupling capacitance and (a)

initial point
Ci is the capacitance of qubit i. Throughout this sec- final point
tion, we will assume a direct capacitive coupling between
qubits of the flux-tunable transmon type, so that g = gq-q

Frequency (GHz)
5.0
and ωqi → ωqi (Φi ). For simplicity, we suppress the ex-
plicit flux dependence of the ωqi ’s and simply refer to the
coupling as g. Equation (103) can be rewritten as

Hqq = −g [σ + − σ − ] ⊗ [σ + − σ − ] , (106)


4.5

and then using the rotating wave approximation again

(i.e. dropping fast rotating terms) we arrive at 0.0 0.1 0.2 ΦiSWAP 0.3 0.4
Magnetic flux, qubit 1 (Φ0)
Hqq = g eiδω12 t σ + σ − + e−iδω12 t σ − σ + , (107) (b) (c)


X

QB1
0
flux 0 τ p p
where we have introduced the notation δω12 = ωq1 − ωq2
and suppressed the explicit tensor product between qubit 4/2 p
subspaces. If we now change the flux of qubit 1 to bring
it into resonance with qubit 2 (ωq1 = ωq2 ), then 3/2

Time, τ

g
Hqq = g σ + σ − + σ − σ + = (σx σx + σy σy ) . (108) 2/2
2
The first part of Eq. (108) shows that a capacitive inter- 1/2
action leads to a swapping of excitations between the two
qubits, giving rise to the ‘swap’ in iSWAP. Moreover, due
to the last part of Eq. (108), this capacitive coupling is ΦiSWAP 0 0.5 1
also sometimes said to give rise to an ‘XY ’ interaction218 . Magnetic flux, qubit 1 (Φ0) Probability
The unitary corresponding to a XY (swap) interaction
is FIG. 15. (a) Spectrum of two transmon qubits (written in the
combined basis as |QB1, QB2i) as the local flux through the
1 0 0 0
 
loop of qubit 1 is increased. Black/dashed lines with arrows
g 0 cos(gt) −i sin(gt) 0 indicate a typical flux trajectory to demonstrate operation of
Uqq (t) = e−i 2 (σx σx +σy σy )t = 
0 −i sin(gt) cos(gt) 0 iSWAP gate. (b) Probability of swapping into the |01i state
0 0 0 1 as a function of time and flux. The pulse sequence corresponds
(109) to preparing |10i and performing a typical iSWAP operation
Since the qubits are tunable in frequency, we can now (for a time τ ). (c) Probabilities of |01i (black) and |10i (gray)
consider the effect of tuning the qubits into resonance at Φ = ΦiSWAP (white dashed line in (b)) as the time spent
for a time t0 = 2gπ
, at the operating point (τ ) is increased. This simulation does
not include any decay effects.
1 0 0 0
 
0 0 0
 
π −i
Uqq = ≡ iSWAP. (110)
2g 0 −i 0 0
0 0 0 1 To elucidate the operating principle behind an iSWAP
implementation we show the spectrum of a flux-
From this result, we see that a capacitive coupling be- tunable qubit using typical transmon-like parameters in
tween qubits turned on for a time t0 (inversely related to Fig. 15(a). The iSWAP is performed at the avoided cross-
the coupling strength in units of radial frequency) leads ing where Φ = ΦiSWAP . By preparing QB1 in state |1i,
to implementing a so called ‘iSWAP’ gate216,217,219–222 , moving into the avoided crossing, waiting there for a time
which acts to swap an excitation between the two qubits, τ (see pulse-sequence in inset in Fig. 15(b)), the excita-
and add a phase of i = eiπ/2 . For completeness, we note tion is swapped back and forth between the two qubits,
that for t00 = 4g
π
the resulting unitary, as shown in Fig. 15(b). In Fig. 15(c), we plot linecuts
of (b) at ΦiSWAP , showing the excitation oscillating back
1 0 0 0
 
√ √ and forth between |01i and |10i with the predicted time
0 1/ 2 −i/ 2 0 √
 
π
Uqq = √ √ ≡ iSWAP, (111) t0 = π/2g. In turn, the frequency of the oscillation can
4g 0 −i/ 2 1/ 2 0 be used to extract the strength of the coupling, t20 = πg .
0 0 0 1
So far we have ignored the role of the single-qubit
is typically
√ referred to as the ‘squareroot-iSWAP’ gate. phases acquired by tuning the qubit frequency. Refer-
The iSWAP gate can be used to generate Bell-like su- ring to the pulse-sequence shown in the top panel of
perposition states, e.g. |01i + i|10i. Fig. 15(a), we see that each qubit will acquire a phase
 33

given by iSWAP gate interspersed with single-qubit rotations was

Z τ used to generate successive XY , XZ and Y Z interactions
θz = dt (ωq − ω(t)) . (112) that lead to an aggregate HHeisenberg Hamiltonian. State-
0 of-the-art operation of the iSWAP gate has also been used
This phase can be conveniently removed either by sub- to demonstrate a ten-qubit GHZ state226 .
sequent application of virtual-Z gates to all following
pulses192 , or by shaping the waveform of the excursion
such that single-qubit phases are exactly cancelled223 . F. The CPHASE two-qubit gate in tunable qubits
Equations (105) and (109) together present a useful re-
sult from a quantum processor design perspective: The In our discussion of the iSWAP gates, we assumed that
operating regime, frequency and time τ of the iSWAP the higher energy levels of the superconducting qubit do
can be calculated (typically simulated) to high precision, not play a role. As we show below, it turns out that for
before any processor fabrication is undertaken. The only the case of transmon qubits (with negative anharmonic-
‘quantum’ parts that enter gqq (and gq-r-q ) are the qubit ity), the higher levels can in fact be utilized to generate
frequencies, ωq1 (Φ1 ) and ωq2 (Φ2 ). If the Josephson en- a the CPHASE gate directly64,227 .
ergies of the qubits are known (which they typically are, Recall from Sec. IV C that the CPHASE gate imple-
from fabrication parameters), then by simulating the ca- ments the following unitary,
pacitances in gqq or gq-r-q , the time τ and the pulseshape
needed to implement an iSWAP can be estimated to 1 0 0 0
 
high precision. Typical values of the coupling strength, 0 1 0 0 
UCPHASE =  (115)
g/(2π), for architectures using the iSWAP gate is 5-40 0 0 1 0
MHz and are often very close to expectations from EM 0 0 0 −1
simulations220,222–224 .
Our goal for the remainder of this section is to show that
the unitary operator of the CPHASE gate appears natu-
2. Applications of the iSWAP gate rally for capacitively coupled transmon superconducting
qubits and review a few of the modern applications of this
gate. We have chosen to include a considerable amount of
The iSWAP cannot generate a CNOT gate by itself.
details for the implementation of this gate, as a means to
Rather, to implement a CNOT gate requires stringing
review some of the issues one has to resolve, to engineer
together two iSWAPs and several single qubit gates218 ,
high quality two-qubit gates.
• Z- π2 X π2 The structure of the matrix in Eq. (64) indicates that
= we need to apply a phase (−1 = eiπ ) to the qubits when-
iSWAP iSWAP ever both are in the excited state |11i. Considering the
X π2 Z π2 Z π2 nature of the XY interaction, which couples |01i ↔ |10i
(113) and leads to the iSWAP gate (see previous section), we
As evident from Eq. (113), the iSWAP gate in general expect avoided level crossings to exist between higher lev-
needs to be used twice to generate a single CNOT, leading els, e.g. |11i ↔ |20i and |11i ↔ |02i. The flux-tunable
to a significant overhead when compiling CNOT–dense implementation of the CPHASE gate relies on this higher-
circuits from iSWAP gates. However, depending on the level avoided crossing.
context, the iSWAP can be used efficiently (i.e. with- To motivate this intuition we plot the spectrum for two
out any two-qubit gate overhead) to mimic the behav- coupled transmon qubits, in Fig. 16(a), including levels
ior of a CNOT. Typically such circuits will not be com- with two excitations, as the local magnetic flux in qubit
pletely equivalent, but will share certain salient features 1 is being tuned. The Hamiltonian for this spectrum,
for specified input states. As an example of this proce- written in the {|00i, |01i, |10i, |11i, |02i, |20i}-basis, is ap-
dure, Neeley et al.221 demonstrated the generation of a proximately given by,
3-qubit Greenberger-Horne-Zeilinger (GHZ) state (which
E00 0 0 0 0 0
 
requires two subsequent CNOTs in the simplest construc-
tion), by using only two iSWAPs in a circuit that cor-  0 E01 g 0 0 0 
 0 0 0 0 
 
rectly generates the 3-qubit GHZ state on the |000i input. g E 10 √ √
H2 excitations =  ,
 0 0 0 √ 2g 2g 

Moreover, the XY –interaction is a powerful tool for cer- E11
tain types of quantum simulation algorithms225 . If one  0 0 0 √2g E02 0 


is interested in digital quantum simulation of spin–like 0 0 0 2g 0 E20

systems, then the XY –interaction can natively simulate (116)
e.g. a Heisenberg interaction, where Enm = Enq1 (Φ1 ) + Em q2
(Φ2 ) and En (Φi ) is the flux-
dependent energy of the i-th level of a transmon52 , and
HHeisenberg = Jx σx σx + Jy σy σy + Jz σz σz . (114)
the √{|02i, |20i} ↔ |11i transitions are scaled by a fac-
This approach to the XY interaction was demonstrated tor 2 due to the higher photon number. In Fig. 16,
by Salathé et al.223 , where repeated application of the we plot the frequencies ωnm = Enm − E00 calculated
 34

(a) where
CPHASE initial point
10 final point Z τ
9
�(τ) θij (`(τ )) = dt ωij [`(t)] (118)
0
8
ω/ 2π (GHz)

is the phase acquired by the state |iji along the trajectory

7
` in (Φ, t)-space during time τ . The movement should be
6 sufficiently slow on a time-scale set by g that the mov-
iSWAP ing state never populates the |20i state, i.e. the move-
5 ment should be adiabatic. In terms of applied flux, the
4
avoided crossing between the |11i ↔ |20i state happens
0.0 0.1 0.2 0.3 0.4 before |10i ↔ |01i (due to the negative anharmonicity of
ΦCPHASE ΦiSWAP the transmons, α ≈ −Ec ) and consequently ` does not
Magnetic flux, qubit 1 (Φ0) take the states through the ΦiSWAP operating point. As
(b)
shown in Fig. 16(b) we can define a parameter (typically
denoted ζ) quantifying the difference in phase acquired
by the |11i relative to the single excitation states,
ω/ 2π (GHz)

�(τ) ζ = (ω11 − (ω01 + ω10 )) . (119)

ζ

The parameter ζ can be thought of as the result (in the

computational space) of the repulsion of |11i due to the
|20i state. If we now choose a trajectory `π , designed so
0.15 ΦCPHASE 0.2 Rτ
that 0 ζ(`π (t))dt = π, then
Magnetic flux, qubit 1 (Φ0)
Z τ
FIG. 16. (a) Spectrum of two coupled transmon qubits (using ζ(t)dt = π = θ11 (`π ) − (θ01 (`π ) + θ10 (`π )) . (120)
typical transmon-like values for Josephson energies and capac- 0
itances) as the local magnetic flux for qubit 1 is varied. The
two lower branches corresponding to |01i and |10i are involved Inserting this expression into Eq. (117) we see that
in the iSWAP gate operation at Φ = ΦiSWAP . The avoided
crossing indicated in the black rectangle is used to implement
1 0 0 0
 
the conditional phase gate (CPHASE), at Φ = ΦCPHASE . Black
line with arrows indicate a typical trajectory used to imple-
0 eiθ01 (`π ) 0 0
=

Uad 0
.
ment a CPHASE gate (starting at the black circle and ending 0 eiθ10 (`π ) 0 
at the gray circle). (b) Zoom in of the |20i ↔ |11i avoided 0 0 0 ei(π+θ01 (`π )+θ10 (`π ))
crossing highlighted in the black box in (a) at Φ = ΦCPHASE . (121)
The parameter ζ quantifies the difference in energy between
|11i and |01i + |10i and `(τ ) is the trajectory in (Φ, t)–space.
After the adiabatic excursion, one can now apply single-
qubit pulses (or use virtual-Z gates) to exactly cancel the
single-qubit phases such that θ10 (`π ) = θ01 (`π ) = 0. This
changes Uad to
from Eq. (116), using standard, symmetric, transmon-
like parameters, as the local magnetic field of qubit 1 is 
1 0 0 0
 
1 0 0 0

increased.
0 1 0 0  0 1 0 0
The result of the higher levels on the computational Uad = = = UCPHASE . (122)
0 0 1 0  0 0 1 0
basis can be understood by considering a concrete ex- 0 0 0 eiπ
0 0 0 −1
ample. By preparing the combined qubit state |11i and
moving slowly towards the avoided crossing between |11i From Eq. (122) it is evident that an adiabatic movement
and |20i at ΦCPHASE , waiting for some time τ and mov- of |11i, followed by single-qubit gates (virtual or real)
ing back (see black line with arrows in Fig. 16(b)), the efficiently implements a CPHASE and, through Eq. (67),
resulting unitary operator in the computational basis is also efficiently implements a CNOT. The CPHASE gate is
given by one of the workhorses of modern superconducting qubit
processesors with gate fidelities & 0.9965,228 .
1 0 0 0
 
One is, of course, free to choose an arbitrary trajec-
0 eiθ01 (`) 0 0  tory `φ that implements the phase e−iφ on the |11i state.
Uad =
0
, (117)
0 e iθ10 (`)
0  Assuming that the single-qubit phases are properly can-
0 0 0 eiθ11 (`) celled, one sees that the arbitrary phase version of the
 35

CPHASE gate (typically denoted CZφ ) can be written as form (based on a Slepian waveform242 ) to parametrize
the trajectory `?π (τ ).
10 0 0
 
01 0 0 
CZφ = 
00 1 0 
00 0 e−iφ
  2. The CPHASE gate for quantum error correction
φ
= exp −i (σz ⊗ σz − σz ⊗ 1 − 1 ⊗ σz ) . (123)
4
Using the approach of Martinis and Geller, Barends
Because of the form of Eq. (123), one can think of the et al. were able to demonstrate a two-qubit gate fidelity
avoided crossing with the higher levels outside the com- FCPHASE = 0.9944 (determined via a technique known
putational subspace as giving rise to an effective σz ⊗ σz as ‘interleaved randomized benchmarking’243–246 ). This
coupling within the computational subspace227 . implementation had a gate time τ = 43 ns and was imple-
An alternative to the adiabatic approach outlined mented with the `?π waveform65 , in an “xmon” device85
above to realize CPHASE is to make a sudden excur- – a transmon with a “+”-shaped capacitor. A two-qubit
sion to the √ΦCPHASE operating point, after waiting a gate fidelity F > 0.99 represents a significant milestone,
time t = π/ 2g, the state will have completed a single not just from a technical and engineering perspective,
Larmor-type rotation from |11i to |02i and back again to but also from a foundational standpoint: The surface
|11i, but in the process, acquired an overall π phase, sim- code (a quantum error correcting code) has a lenient
ilar to the iSWAP gate, but in the {|11i, |20i} subspace54 . fault-tolerance threshold of ∼ 1%247–249 . This means,
In fact, such excursions near or through avoided crossings roughly speaking, that if the underlying operations on
leading to adiabatic and non-adiabatic transitions have the qubits have fidelities F > 0.99, then by adding
been studied extensively in the context of interferometry, more qubits to the circuit (and correctly implementing
cooling, spectroscopy, and quantum control119,229–238 . the fault-tolerant quantum error correction protocol) the
The remainder of this subsection is devoted to an overall error-rate can be reduced, and one can in principle
overview of some of the recent advances and demonstra- perform arbitrarily long quantum computations, with-
tions using the CPHASE gate since its first demonstra- out errors spreading uncontrollably and corrupting the
tion in 2009 where it was used to generate Bell-states calculation. Because of its relatively lenient threshold
and demonstrate two-qubit algorithms64 . under circuit noise (compared to e.g., Steane or Shor
codes178,250,251 ) and its use of solely nearest-neighbor
coupling, the surface code is one of the most promising
1. Trajectory design for the CPHASE gate quantum error correction codes for medium-to-large scale
quantum computing in solid state systems247 . Therefore,
The (adiabatic) implemention of UCPHASE outlined surpassing the fault-tolerance threshold using CPHASE
above assumed that the trajectory `π was completely represents a significant milestone for the field252 . More-
adiabiatic and that the |11i state never left the compu- over, practical blueprints for implementing scalable sub-
tational subspace. Since the fidelity of gates is bounded cells of the surface code using the CPHASE as the funda-
from above by the coherence times of the qubits, short mental two-qubit gate have also been proposed71 as well
gate times are desirable239 . This presents a tension for as in-situ calibration protocols for large-scale systems op-
optimally operating the CPHASE gate – fast operation erating with CPHASE253 . For a full review of the pros and
in conjunction with the need for adiabatic operation. A cons of various quantum error correcting codes we refer
relevant question is then: what is the optimal trajectory the interested reader to e.g. an introductory review arti-
`?π that implements the necessary phase as fast as pos- cle Ref. 254, or any of the excellent textbooks and more
sible, with as little leakage as possible, for a given size detailed review articles in Refs. 178, 180, 251, 254–257.
of the avoided crossing between |11i and |20i? Given a Returning to the CPHASE gate, numerical optimiza-
typical coupling rate g/2π ≈ 20 MHz (as discussed in tion of `?π was demonstrated by Kelly et al.228 using the
Sec. IV E), one expects a heuristic lower time limit to be interleaved randomized benchmarking sequence fidelity
2π/g ≈ 50 ns (stronger coupling of course leads to shorter as a cost function to push a native implementation of
gate times, but will limit the on/off ratio of the gate). `?π with a fidelity F = 0.984 up to F = 0.993, sur-
Traditional optimal control of adiabatic movement as- passing the surface code fault tolerance threshold. In
sumes the movement is through the avoided crossing (see the same work that demonstrated FCPHASE = 0.9944,
e.g. Ref. 240), but the trajectory `π moves close to and Barends et al.65 used the CPHASE gate
√to generate
then back from the avoided crossing. This modification GHZ states, |GHZi = |0i⊗N + |1i⊗N / 2, of up to
to the adiabatic movement protocol was addressed by N = 5 qubits, with a fidelity for the N = 5 state of
Martinis and Geller241 , specifically in the context of er- F = Tr (ρideal ρN =5 ) = 0.817. The protocol for generat-
rors for a CPHASE gate implemention. The authors show ing the GHZ state with N = 2 and N = 3 from CPHASE
that non-adiabatic errors can be minimal for gate times was originally demonstrated by DiCarlo et al.54,64 . The
only slightly longer than 2π/g using an optimal wave- textbook route to generating the N = 2 GHZ state, |Φ+ i
 36
 
(a Bell state) from the all-zero input is exp −i φ2 σz ⊗ σz unitary can be generated via

(124)

|0i H • 
 = |Φ i ,
+
|0i Xπ Aπ
UZZ (φ) = CZφ CZφ
An equivalent circuit using CPHASE and native single- Aπ Xπ
qubit gates in superconducting qubits is: (129)
where Aπ ∈ {Xπ , Yπ } is used to allow for small and neg-
|0i Yπ/2 • . (125) ative angles. Finally, for completeness, we mention an
alternative approach to creating UZZ , given by42,264
|0i Y−π/2 • Yπ/2
• • ,
By repeating the operation inside the dashed box UZZ (φ) = (130)
on additional qubits, an N -qubit GHZ state can be Zφ/2
generated65 . Since the demonstration of the N = 5 GHZ
state using the CPHASE gate, the gate has been deployed
which has the benefit of relying on CPHASE (through
to demonstrate several important aspects of quantum in-
the CNOTs), and the angle can be controlled using the
formation processing using superconducting qubits. A
single-qubit Z gates. We refer the interested reader to
nine-qubit implementation of the five-qubit repetition
two reviews on quantum simulations, see e.g. Refs. 265
code (five data qubits + four syndrome qubits)254 was
and 266.
demonstrated, and the error suppression factor of a sin-
gle logical quantum bit was shown to increase as the en- The CPHASE gate has also been used in a vari-
coding was changed from three data qubits to five data ety of other contexts, e.g., for calculating the dissoci-
qubits66 . Similarly, in a five qubit processor the three- ation of diatomic hydrogen (H2 ) using the variational
qubit repetition code with artificially injected errors was quantum eigensolver method267 , for feed-forward based
demonstrated258 , building on earlier results utilizing a teleportation experiments268,269 , as well as initial steps
combination of iSWAP and CPHASE gates to perform towards demonstrating quantum supremacy270 and a
parallelized stabilizer readout259 . 2 × 2 implementation of the Harrow-Hassidim-Lloyd
algorithm271,272 . In the field of hybrid semiconducting
nanowire/superconducting qubits (known as the “gate-
mon” approach273–275 ), where the qubit frequency is
3. Quantum simulation and algorithm demonstrations
using CPHASE
modified by electrostatically changing the density of car-
riers in a semiconducting region with proximity-induced
superconductivity, the CPHASE gate was also demon-
As an example of the utility of the CPHASE gate, we strated between two nanowire qubits276 .
briefly discuss a particular demonstration of a digital
quantum simulation. In this context, the CPHASE gate One may worry that operating a qubit by moving its
has been utilized to simulate a two-site Hubbard model frequency can lead to overlap with frequencies already
with four fermionic modes, using four qubits260 . Using used by other qubits, in a system with multiple qubits.
the Jordan-Wigner transformation261,262 , it is possible to This issue is known as frequency crowding. While the
map fermionic operators onto Pauli spin matrices263 . As use of asymmetric transmons (with two sweet spots in
shown in Ref.260 a Hubbard model with two fermionic the range [−Φ0 , +Φ0 ], recall Fig. 2(c)) may help allevi-
modes, whose Hamiltonian is given by ate some frequency crowding issues, a more long-term
strategy is needed. One way to circumvent the problem
is to utilize on/off tunable coupling schemes, in which
HHubbard, two mode = −t(b†1 b2 + b†2 b1 ) + U b†1 b1 b†2 b2 (126)
qubits can exchange energy only if a coupler activates
the interaction63,104 . To address this issue in the context
can be written in terms of Pauli operators as,
of the CPHASE gate, Chen et al.104 demonstrated a de-
t vice (named “the gmon”) where the qubit interaction can
H= (σx ⊗ σx + σy ⊗ σy ) (127) be tuned with an on/off ratio on the order of 1000, and a
2
U CPHASE gate fidelity of F = 0.9907 was demonstrated.
+ (σz ⊗ σz + σz ⊗ 1 + 1 ⊗ σz ) , (128) This concludes the introduction to the physics and op-
4
eration of the CPHASE gate in its native form. In the
where U is the repulsion energy and t is the hopping remainder of this section we will introduce a few of the
strength. Similar to the Heisenberg interaction discussed microwave-only gates that have been demonstrated in an
briefly in Sec. IV E, it is now a question of producing effort to sidestep the need for local tunability (and the
ασi ⊗ σi -type interactions, where the prefactor α can be resulting increased sensitivity to noise) as required by the
tuned. Using the CZφ version of CPHASE, a UZZ (φ) = iSWAP and CPHASE gate.
 37

G. Two-qubit gates using only microwaves (a) ~ (b)

ω 2 g
g2
+
One common (potential) drawback for the iSWAP 2Δ12
~ ~
and CPHASE gates is that their operation requires flux- ω1 ω 2 g2
-
2Δ12
tunable qubits. Introducing a new control knob, such as
flux control, in turn also introduces a new noise channel
for the system. Furthermore, the need for flux-tunability QB1 coupler QB2
increases the sensitivity of the devices to flux noise by (c) (d)
QB1=
tuning the qubits from their “sweet spots”, increases the 1 π

Expectation value, QB2

dephasing rate. From this perspective, one could envi-

angle, QB2 (rad)

sion using all-microwave-based gates to remedy these is- 0
sues. To this end, the cross-resonance (“CR”) gate was -1
developed for operating fixed-frequency superconducting QB1=
1
qubits277–279 , which typically feature longer lifetimes and
reduced sensitivity to flux noise. 0

-1 0
0 100 200 0 100 200
1. The operational principle of the CR gate Time (ns) Time (ns)

To elucidate the operation of the CR gate, we briefly FIG. 17. (a) Schematic circuit diagram of two fixed frequency
revisit the driving Hamiltonian derived in Sec. IV D. transmons coupled through a resonator yielding an overall
There, we considered only a single qubit. However, if coupling coefficient g. Qubit 1 driven at the frequency of
one extends this formalism to two qubits, see Fig. 17(a) qubit 2 leads to the CR gate. (b) Schematic level diagram
denoting the frequency difference by ∆12 = ωq1 −ωq2 and of the always-on coupling leading to dressed states |01i e and
the coupling by g  ∆12 , and performing a Schrieffer- |10i
e with ∆12 = ω1 − ω2 . (c) Simulations of the expectation
Wolff transformation to go to the dressed state picture, values of hσz i and hσy i for qubit 2 as a drive at the frequency
the driving Hamiltonians for qubit 1 and 2 become278,280 of qubit 2 is applied to qubit 1. Upper panel shows regular
Rabi oscillations when qubit 1 is in the |0i state. Bottom
Hd,1 = ΩVd1 (t) σx ⊗ 1 + ν1− 1 ⊗ σx + µ− panel shows a modified Rabi frequency when qubit 1 is in |1i


1 σz ⊗ σx
state, in accordance with Eq. (135). (d) Difference in angle
(131) in the (z, y) plane as a function of length of the applied drive
Hd,2 = ΩVd2 (t) 1 ⊗ σx + ν2 σx ⊗ 1 + µ2 σx ⊗ σz
+ +


to qubit 1. At approximately 200 ns π-phase shift has been
(132) acquired.

where
g αi
µ±
i =± (133)
∆12 (αi ∓ ∆12 )
This strategy was first demonstrated using flux-
g ∓∆12
νi± =± (134) tunable transmons in Ref.282, where a Bell state with
∆12 (αi ∓ ∆12 ) fidelity Fbell = hΦ+ |ρ|Φ+ i = 0.90 was achieved. Using
and ΩVdi (t) is the driving for qubit i. From Eq. (131), quantum process tomography the gate fidelity was found
it is evident that if we drive qubit 1 at the frequency of to be FQPT = 0.81. By moving to fixed-frequency qubits
qubit 2, then to qubit 2, this will look like a combination with increased lifetimes, the gate fidelity was increased
of ν1− 1 ⊗ σx and µ− to FQPT = 0.98 (with subtraction of state initialization
1 σz ⊗ σx . This means that the Rabi
oscillations of qubit 2 will have a frequency given by and measurement errors)281 . For completeness, we note
that due to the form of the last term in Eq. (131), the
ΩRabi −
QB2 = ΩVd1 ν1 + z1 µ1 ,
−
(135) CR gate is also sometimes denoted the ZXθ gate. The


unitary matrix representaion of the CRθ gate is
where z1 = hσz 1i, and z1 depends on the state of qubit
1. This effect is demonstrated in Fig. 17(c), where a
simulated drive is applied to qubit 1 while the resulting i
UCRθ = e− 2 θσz ⊗σx
Rabi oscillations in qubit 2 are recorded. We have used
cos θ2 −i sin θ2 0 0
 
typical fixed-frequency transmon parameters from exper-
−i sin θ cos θ 0 0 
iments, and we have included a spurious cross-talk term = 2 2 (136)
 0 0 cos 2 i sin θ2 
θ

η = 0.03.246,281 . In Fig. 17(d), we plot the difference in
angle in the (z, y) plane acquired by qubit 2 for different 0 0 i sin θ2 cos θ2
initializations of qubit 1, ∆φ = φzy zy
|00i − φ|10i . For this
particular choice of parameters, the cross-resonance gate
achieves a π-phase shift in ≈ 200 ns. where θ = −µ−
1 ΩVd1 (t), which can be used to generate a
 38

CNOT with the addition of only single-qubit gates, gate times (τCPHASE = 30 – 60 ns and τCR = 300 – 400
ns), which to a large extent accounts for the observed
• Z π2 CR− π2 , CR gate fidelities. The time scale for CR operation is set
= (137) by the frequency detuning, the anharmonicity, and the
X π2 • coupling strength, through Eq. (133). This has the un-
fortunate drawback that if qubits do not have intended
frequencies (due to fabrication variation), it will be im-
up to a phase eiπ/4 . mediately manifest as longer gate times, and in turn, re-
duced gate fidelity. As fabrication techniques are becom-
ing more sophisticated and reliable, this problem may be
2. Improvements to the CR gate and quantum error of reduced importance. However, since the coupling in
correction experiments using CR
the CR scheme is always on, there is an inherent tension
between well-isolated qubits for high-fidelity single-qubit
Since qubit 1 is being driven off-resonance, an ac-Stark operations, and coupling qubits, for fast/high-fidelity two
shift will add a term ∝ σz 1 to the driving Hamiltonian of qubit gates.
qubit 1. The effect of both the spurious ac Stark shift and
the direct ν1− 1σx single-qubit rotations was studied in
Ref.246. By modifying the original CR protocol to effec-
3. Quantum simulation and algorithm demonstrations
tively “echo away” the two unwanted contributions from with the CR gate
the σz 1 and 1σx terms, the fidelity of the CR gate was
improved to FCR = 0.8799246 , using quantum process
tomography. Using interleaved randomized benchmark- Since the form of the CR Hamiltonian (σz ⊗ σx ) is
ing of this improved “echo-CR”-gate (eCR− π2 ), a gate not a (σx ⊗ σx + σy ⊗ σy )-type interaction (leading to
fidelity of FeCR− π = 0.9347 was achieved. This gate iSWAP gate) nor is it an the effective (σz ⊗ σz )-type
2 (leading to CPHASE gate), one could question its appli-
implementation was used to demonstrate two-qubit par-
cability to quantum-simulation-type experiments, which
ity measurements in a three-qubit device283 , as well as
often involves terms of the form σi ⊗ σi . However, by de-
detecting bit-flip and phase-flip errors in a Bell state en-
veloping a variational quantum eigensolver routine that
coded in a four-qubit device284 , with gate fidelities from
efficiently generates entangled trial states using just the
interleaved randomized benchmarking in the range 0.94
CR interaction, Kandala et al.288 calculated the ground-
to 0.96. Using a similar device, but with five qubits,
state energy for H2 , LiH, and BeH2 . This experiment was
weight-four parity measurement of the forms ZZZZ and
performed on six fixed-frequency qubits, and it employed
XXXX were demonstrated285 , where the crosstalk to
a technique for compact encoding of the Hamiltonians
qubits not involved in the CR gates was studied, leading
corresponding to each molecule289 . As of this writing,
to the development of a four pulse eCR4-pulse scheme.
this experiment represents the largest molecule for which
Based on improvements in the analysis of the Hamil-
the ground state has been found using a purely quantum
tonian describing the CR drive, Sheldon et al.203 sub-
processing approach.
sequently demonstrated a version of the CR which re-
The CR gate is also the native two-qubit gate available
duced the gate time to τ = 160 ns and added an ac-
on the IBM Quantum Experience quantum processor290 ,
tive cancellation tone to the eCR previously developed.
which is accesible online. Using the IBM Quantum Expe-
Using this “active cancellation echo CR” (aceCR), the
rience processor, Takita et al.291 demonstrated an imple-
fidelity was increased to FaceCR− π = 0.991, measured
2 mentation of a two-logical-qubit (four physical qubit) er-
with interleaved randomized benchmarking. The same ror detection code292 . The implementation was inspired
sequence without active cancellation on the same qubits by the proposal of Gottesman293 , which proposed a min-
yielded FeCR− π = 0.948. The interested reader may con- imal experiment to claim observation of fault-tolerant
2
sult the followup theoretical work286 with more details encodings255 , using a four qubit error detection code in
on the effective Hamiltonian models. Other approaches a five qubit setup. Due to constraints on the connec-
to fast, high-fidelity cross-resonance gates have also been tivity, the work by Takita et al. demonstrated a mod-
proposed287 . This series of improvements to the original ified version of the Gottesman encoding, in which two
cross-resonance implementation has increased the gate fi- logical qubits are initialized, but only one of them in a
delity to beyond the threshold for fault-tolerance in a sur- fault-tolerant manner. By artificially injecting an error
face code, with similar quality to the CPHASE gate. Al- in the state preparation circuit, the authors demonstrate
though improvements should still be made, with the ad- that the probability of correctly preparing a fault tolerant
vent of the CR gate, superconducting qubit based quan- state is greater than the probability of correctly prepar-
tum computing platforms now offer two entangling two- ing a non-fault-tolerant qubit. This behavior is consistent
qubit gates that can be used for implementing surface- with expectations for how fault-tolerant encodings work.
code based error correction schemes. Simultaneously, Vuillot294 also used the IBM Quantum
In the initial experiments using CR gates, the gate Experience machine to study fault-tolerant schemes en-
times were significantly longer than the typical CPHASE coded in that connectivity.
 39

Beyond the applications to error-correction and error- The two unitaries UZZ and UIZ−ZI only contain terms
detection, the cross-resonance gate has also been em- that commute with UbSWAP (θ, φ), and their effects can be
ployed in early demonstrations of quantum advantages offset in post-processing280 . In Eq. (139), φ is the phase
in machine learning. Risté et al.295 studied the so-called of the drive relative to the single-qubit drive pulses, and
“learning parity with noise” problem, in which one at- θ = ΩB t with
tempts to learn a bit-string k by querying an oracle func-
−2gΩ2 −gγαΣ + γ 2 α2 (α1 + ∆12 ) + α1 (α2 − ∆12 )


tion f (D, k) = D · k mod 2 with a user-input bit-string
ΩB = ,
D. In a first implementation of this problem, the authors (α1 + ∆12 )(α2 − ∆12 )∆212
show that for a specific instance of the bit-string k = 11, (140)
a learner with access to quantum operations needs fewer where Ω is the amplitude of the drive, γ is a dimen-
queries to the function f . However, by extending the sionless parameter quantifying the coupling coefficient
model of learning parity with noise, the authors demon- of the drive to qubit 2 in units of coupling strength
strated a consistent advantage of the learner with access to qubit 1, and αΣ = α1 + α2 . Explicit derivations
to quantum operations295 . leading to Eq. (138) can be found in the supplement
The CR gate was also used to demonstrate the imple- of Ref.299. By applying UbSWAP for a time that yields
mentation of a supervised learning algorithm where the θ = π/2, and with φ = 0, the resulting gate is denoted
feature space is encoded as quantum data on the Bloch bSWAP and can act as the entangling gate (together with
sphere264 . In typical supervised learning, an algorithm single-qubit gates) that forms a universal gate set. More-
is exposed to a training set of labeled data, and is subse- over, the power of the bSWAP becomes apparent when
quently asked to classify a new, unlabeled set of data296 . one applies it for the time that yields θ = π/4, which
In the support vector machine (SVM) approach to such from the ground state |00i directly produces the entan-
problems, the data is then mapped non-linearly onto the gled Bell
√ state |00i + e |11i. In line with
iφ
√ the defini-
so-called “feature space”, in which the trained algorithm tion of iSWAP, this gate is denoted the bSWAP. In
has constructed a separating hyperplane to classify the the work by Poletto et al.299 , the fidelity of the bSWAP
data. While a full “quantum Support Vector Machine” gate was FbSWAP = 0.9 (determined from quantum pro-
proposal exists, the algorithm assumes that the data are cess tomography). The main source of error was the in-
already present in a coherent superposition297 . Instead, creased dephasing during the relatively long high-power
Havlicek et al.264 proposed, and demonstrated, that map- pulse needed to drive the |00i ↔ |11i transition. The
ping the classical data non-linearly onto the Bloch sphere bSWAP gate can be viewed as the superconducting qubit
can also be utilized to provide a quantum advantage. analogue of the Mølmer–Sørensen gate300 . In Fig. 18, we
For a wider discussion of the important role of quantum outline the level diagram of two coupled qubits, along
data in many quantum machine learning algorithms, the with the higher levels of the qubits. The arrows indicate
reader is referred to Ref.298 which coupled states are utilized to implement the cor-
responding gate. As an application of the bSWAP gate,
Colless et al.301 used this gate to calculate energies of
4. Other microwave-only gates: bSWAP, MAP, and RIP the excited states of a H2 molecule using a two-qubit
transmon processor301 .
The CR gate (as outlined above) is not the only all- Another all-microwave gate is the so-called
microwave two-qubit gate available. In particular, the ”microwave-activated CPHASE” (or ‘MAP’ for short)70 .
bSWAP gate299 is an interesting alternative. The bSWAP The MAP gate is in spirit similar to the CPHASE gate,
gate directly drives the |00i ↔ |11i transition, made pos- where noncomputational states are used to impact a
sible by interactions with the higher levels of the qubit, conditional phase inside the computational subspace.
see Fig. 18. Usually, the matrix element for such a In contrast to CPHASE, the MAP gate is implemented
transition is small (3rd order in the coupling strength), without tuning individual qubit frequencies. Rather,
but if the detuning between the qubits is equal to the the canonical implementation of this gate comprises two
anharmonicity, the transition rate is enhanced. Apply- fixed-frequency qubits, where the frequencies are care-
ing a sequence of Schrieffer-Wolff transformations to the fully designed (and fabricated), such that the |12i and
coupled-qubit system, and using a carefully chosen drive |03i levels are resonant. This leads to a splitting of the
frequency (close to the midpoint of ωq1 and ωq2 ), it can otherwise degenerate |02i ↔ |01i, and |12i ↔ |11i tran-
be shown280 that the drive gives rise to a unitary operator sitions. By driving near resonance with the |n2i ↔ |n1i
U = UbSWAP (θ, φ)UZZ UIZ−ZI (138) transition, an effective σz ⊗ σz interaction is generated.
In a setup comprising two fixed-frequency qubits, the
with MAP gate was used to implement the unitary
cos θ 0 0 −ie−i2φ sin θ
 
h π i
0 1 0 0 UMAP = exp −i σz ⊗ σz , (141)
UbSWAP (θ, φ) =  4
 
,
 0 0 1 0 
−ie−i2φ sin θ 0 0 cos θ with a gate fidelity FMAP = 0.87 (determined via quan-
(139) tum process tomography) in a time τMAP = 514 ns70 .
 40

bus, χ is the dispersive shift, and ∆cd is the detuning of

the drive (d) from the cavity (c). By choosing θ̇t = π/4, it
is possible to implement the CPHASE gate. The power of
Ea the RIP gate lies in its capability to accommodate large
differences in qubit frequencies. To demonstrate this,
Paik et al.304 performed two-qubit randomized bench-
marking between pairs of qubits in a four-qubit device
Eb with frequency differences spanning 0.38 GHz to 1.8 GHz,
all with fidelities in the range 0.96-0.98 and gate times in
the range 285 to 760 ns.

H. Gate implementations with tunable coupling

computational subspace
Finally, we briefly review tunable coupling architec-
tures, which have recently emerged as a promising al-
FIG. 18. Schematic of the level structure of two coupled ternative. The idea is to engineer an effective qubit-
qubits (including higher levels) with indication of the tran- qubit coupling ge that is tunable (typically by apply-
sitions utilized in the iSWAP, bSWAP, CPHASE and MAP ing a flux), and such gates are referred to as paramet-
gate. See text for details. Figure inspired from Ref. 70.
ric gates. This can be implemented in two different
ways: (i) The coupling strength between two qubits is
tuned by a flux, g → g(Φ(t))199,208,305–307 , or (ii) the
As the number of qubits in a system increase, one draw- resonant frequency of the coupling element is modified
back of this gate is the need for a precise matching of ωcoupler → ωcoupler (Φ(t))91,107,308–312 , with a fixed g,
higher energy levels across multiple qubits, while simul- leading to an effective time-dependent coupling param-
taneously avoiding spurious couplings to other modes in eter. When the tunable coupling element is driven at
the system. frequencies corresponding to the detuning of the qubits
The CR, bSWAP and MAP gates all have quite strin- from the coupler, an entangling gate can be implemented.
gent requirements on the spectral landscape of the qubits In a setup of type (ii), an implementation of the
in order to obtain fast, efficient gate operation. To iSWAP gate was demonstrated by parametrically driv-
address this issue, another all-microwave gate was de- ing a flux-tunable coupler between two fixed-frequency
veloped, the so-called “resonator induced phase gate” qubits63 , yielding a fidelity FiSWAP = 0.9823 (using inter-
(“RIP”)302,303 . The RIP gate operates by coupling two leaved randomized benchmarking) in a time τ = 183 ns.
fixed-frequency qubits to a bus cavity, from which they Similarly, the bSWAP (and iSWAP) gates were recently
are far detuned. By adiabatically applying and removing demonstrated, using a flux-tunable transmon connect-
an off-resonant pulse to the cavity, the system undergoes ing two fixed-frequency transmons. Driving the flux
a closed loop in phase space, after which the cavity is through the tunable qubit at the sum frequency of the
left unchanged, but the qubits acquire a state-dependent fixed-frequency transmons results in the bSWAP105 gate.
phase. By a careful choice of the amplitude and detuning This parametrically driven approach is generally sig-
of the pulse, and taking into account the dispersive shift nificantly faster than implementations relying solely on
of the cavity, a CPHASE gate can be implemented on the fixed-frequency qubits.
two qubits. This effect was experimentally demonstrated A hybrid approach, in which a combination of tunable
by Paik et al.304 in a 3D transmon system55 , where four and fixed-frequency qubits is used, was recently demon-
qubits are coupled to the same bus. In this setup, the strated for both iSWAP and CPHASE gates67,106,212 .
RIP gate operation results in unitaries with weight on all This scheme has no added tunable qubits (or resonators)
four qubits simultaenously. In order to isolate just the acting as the coupling element, but rather, relies solely
desired two-qubit coupling terms, Paik et al. developed on an always-on capacitive coupling between the qubits,
a “refocused” RIP (rRIP) gate that implements and the effective coupling is roughly half that of the
always-on coupling. The operational principle here is
UrRIP = exp −iθ̇σz ⊗ σz t , (142)
 
to modulate the frequency of the tunable qubit (using
local flux control) at the transition frequency correpond-
where the coupling rate (for an unmodulated drive) scales
ing to |01i ↔ |10i for iSWAP and |11i ↔ |02i for
as
CPHASE. Using interleaved randomized benchmarking,
the authors demonstrated FiSWAP = 0.94 (τ = 150 ns),
 2
|ΩVd | χ
θ̇ ∝ , (143) and FCPHASE
02
= 0.93 (τ = 210 ns) and FCPHASE
20
= 0.88
2∆cd ∆cd
| {z } (τ = 290 ns), showing a slight asymmetry in the direc-
n̄
tion in which the CPHASE is applied. This hybrid tech-
where n̄ denotes the average number of photons in the nique was used in Ref.67 to demonstrate a four-qubit
 41

GHZ state with fidelity F4 qubit GHZ = 0.79 (using state Jaynes-Cummings Hamiltonian315–317 , previously intro-
tomography). Finally, this gate-architecture was used to duced in Sec. II,
demonstrate a hybrid quantum/classical implementation
of an unsupervised learning task (determining clustering
1
 
of data), using nineteen qubits and supplemented by a ωq
HJC = ωr a† a + + σz + g σ+ a + σ− a† , (144)


classical computer as part of the minimization loop313 . 2 2

where ωr and ωq denote the resonator and qubit frequen-

cies, respectively, and g is the transverse qubit-resonator
V. QUBIT READOUT
coupling rate. The operators σ+ and σ− represent the
processes of exciting and de-exciting the qubit, respec-
The ability to perform fast and reliable (high fidelity) tively.
readout of the qubit states is an important cornerstone In the limit when the detuning between the qubit and
of any quantum processor3 . the resonator is small compared with their coupling rate,
In this section, we give a brief introduction to how i.e. ∆ = |ωq − ωr |  g, the energy levels of the two
readout is performed on superconducting qubits. We systems√hybridize and a vacuum Rabi splitting of fre-
start by reviewing the fundamental theory behind disper- quency ng/π opens up, where n = 1, 2, 3... denotes the
sive readout – the most common readout technique used resonator mode. In this regime, excitations are coher-
today in the circuit QED architecture – in which each ently swapped between the two systems. Although use-
qubit is coupled to a readout resonator. In the dispersive ful for certain two-qubit gate operations, recall Sec. IV E,
regime, i.e. when the qubit is detuned from the resonator such transverse interactions change the qubit state (since
frequency, the qubit induces a state-dependent frequency energy is directly exchanged between the resonator and
shift of the resonator from which the qubit state can be the qubit) and is therefore not desired in the context of
inferred by interrogating the resonator. quantum non-demolition (QND) readout, in which the
Dispersive readout allows us to map the quantum de- outcome of the quantum measurement is not altered in
gree of freedom of the qubit onto the classical response of the act of reading out the system.
the linear resonator, thus transforming the readout opti- In the dispersive limit, i.e., when the qubit is far de-
mization process into obtaining the best signal-to-noise tuned from the resonator compared with their coupling
ratio (SNR) of the microwave signal used to probe the rate g and the resonator linewidth κ, ∆  g, κ, there
resonator. is no longer a direct exchange of energy between the
We then provide guidance on how to optimize system two systems. Instead, the qubit and resonator push
parameters to perform high-fidelity, single-shot readout. each others’ frequencies. To see this, the Hamiltonian
After choosing parameters, such as resonant frequencies can be approximated using second-order perturbation
and coupling rates, we address the filter and amplifier theory215,318 in terms of g/∆, taken in the limit of few
circuitry positioned in-between the qubit plane and the photons in the resonator. This is known as the disper-
data aquisition hardware outside of the dilution refrid- sive approximation, after which the Hamiltonian takes
erator. On this note, we review the basic principles of the form
Purcell filters as well as parametric amplifiers, both of
which are necessary to obtain fast, high-fidelity readout
1
  
in scaled-up quantum processors. ω
eq
Hdisp = ωr + χσz a† a + + σz , (145)
2 2

where χ = g 2 /∆ is the qubit-state dependent frequency

A. Dispersive readout
shift, a so-called dispersive shift, see Fig. 19(b), allow-
ing us to distinguish the two qubit states. This is an
A quantum measurement can be described as an asymptotically longitudinal interaction, yielding a QND
entanglement of the qubit degree of freedom with a measurement. Note that, in addition, the qubit fre-
“pointer variable” of a measurement probe with a quan- quency also picks up a Lamb shift, ω eq = ωq + g 2 /∆, in-
tum Hamiltonian314 , followed by classical measurement duced by the vacuum fluctuations in the resonator. Also
of the probe. In circuit QED, the qubit (the quantum sys- note that the dispersive Hamiltonian in Eq. (145) is de-
tem) is entangled with an observable of a superconduct- rived for a two-level atom¶ . Taking the second excited
ing resonator (the probe), see Fig. 19(a), allowing us to state into account and introducing the anharmonicity
gain information about the qubit state by interrogating
the resonator – rather than directly interacting with the
qubit. Therefore, the optimization of the readout per-
formance is translated to maximizing the signal-to-noise ¶ Inreality, superconducting qubits, just like natural atoms, have
ratio of a microwave probe tone sent to the resonator, higher energy levels. These higher levels are outside of the com-
while minimizing the unwanted back-action on the qubit. putational subspace, but need to be taken into account for most
The qubit-resonator interaction is described by the qubit simulations to get accurate predictions of frequency shifts.
 42

(a) SIGNAL CREATION SIGNAL DETECTION

(a) /2 = 0 /2 = EC/h
ωIF 0.2
g/2 = 50 MHz

Disp. Shift, ,χ/2 (GHz)

g/2 = 100 MHz
0.1
AWG ωRF ωLO
0.0
Straddling
300 K CRYOSTAT -0.1 regime

3K
10 mK Para. Amp. -0.2
-0.4 -0.2 0.0 0.2 0.4
Frequency detuning, /2 (GHz)
(b) 0

Disp. Shift, ,χ/2 (MHz)

-2

Purcell filter -4
g
κ -6
Qubit

Res. -8 g/2 = 50 MHz

g/2 = 100 MHz
-10
-2.0 -1.5 -1.0 -0.5 0.0
CHIP
Frequency detuning, /2 (GHz)

FIG. 20. (a) Dispersive frequency shift χ/2π as a function

(b) (c)
of qubit-resonator detuning ∆/2π, according to Eq. (146),
1.0
|S11| (a.u.)

for a transmon qubit with anharmonicity α/2π = −EC /h =

κ/2
|S11|
−300 MHz, for qubit-resonator coupling rates g/2π = 50 MHz
0.5
2χ/2 θ
(blue) and g/2π = 100 MHz (red). The two vertical asymp-
totes at ∆/2π = 0 and ∆/2π = EC divides the dispersive
0.0
0 shift into three regimes; For ∆/2π < 0, and ∆/2π > EC /h,
the dispersive shift is negative and χ/2π → 0− as ∆ → ±∞.
θ (rad)

- For 0 < ∆/2π < EC /2π, the dispersive shift χ/2π > 0. This
is called the straddling regime 52 . (b) Zoomed-in plot for neg-
-2 ative qubit-resonator detuning, the most commonly used op-
-2 -1 0 1 2
Frequency, ωRF - ωr (a.u) erating regime.

FIG. 19. (a) Simplified schematic of a representative exper-

imental setup used for dispersive qubit readout. The res- which for a transmon qubit with α < 0 implies that the
onator probe tone is generated, shaped and timed using an dispersive shift will depend on the detuning. This ef-
arbitrary waveform generator (AWG), and sent down into the fect is plotted in Fig. 20(a), where the second energy
cryostat. The reflected signal S11 is amplified, first in a para- level manifests itself as a second vertical asymptote at
metric amplifier and then in a low-noise HEMT amplifier,
∆/2π = EC /h. It is also worth noting that for qubit
before it is downconverted using heterodyne mixing and fi-
nally sampled in a digitizer. (b) Reflected magnitude |S11 | modalities with positive anharmonicity, e.g. flux qubits,
and phase θ response of the resonator with linewidth κ, when the dispersive shift will also shift the sign62 .
the qubit is in its ground state |0i (blue) and excited state |1i In the small photon-number limit, the interaction term
(red), separated with a frequency 2χ/2π. (c) Corresponding of the Hamiltonian in Eq. (145) commutes with the qubit
complex plane representation, where each point is composed observable‖ , σz , resulting in a QND measurement314 .
of the in-plane Re[S11 ] and quadrature Im[S11 ] components. This is an important condition for many applications in
The highest state discrimination is obtained when probing the quantum information processing.
resonator just in-between the two resonances, (dashed line in In the case when the resonator photon number n = a† a
(b)), thus maximizing the distance between the states.
exceeds a critical photon number nc ≡ ∆2 /(4g 2 ), the dis-
persive Hamiltonian in Eq. (145) is no longer a valid
approximation215,319,320 . Therefore, the critical pho-
ton number sets an upper bound for the power level
α = ωq1→2 − ωq0→1 modifies the expression for the dis-
persive shift:

g2 1 ‖ This
 
χ12 commutation is approximate and has an asyptotic dependence
χ = χ01 + = − 01 , (146) on the qubit-resonator detuning
2 ∆ 1 + ∆/α
 43

of the resonator probe signal to maintain (an approx- LPF ADC DSP
imate) QND measurement∗∗ . This limitation could be
lifted by implementing a pure (and not only approxi-
mate) QND readout using a manifestly longitudinal cou-
pling between a qubit and the resonator. Several groups
IF
RF LO

are currenly pursuing the implementation of longitudinal

RF LO
readout, in which QND readout could be performed even
with larger number of resonator photons, thus improving
the SNR108,322,323 . RF
IF
LO

We can also interpret the dispersive qubit-resonator

interaction in another way; by rearranging the terms in
Eq. (145), we can equivalently write
LPF ADC DSP

1 1 g2 2g 2 †
   
Hdisp = ωr a a +
†
+ ωq + + a a σz , FIG. 21. Schematic of an I-Q mixer. A readout pulse at
2 2 ∆
|{z} |∆{z } frequency ωRO enters the RF port, where it is equally split
Lamb shift ac-Stark shift into two paths. A local oscillator at frequency ωLO enters
(147) the LO port, where it is equally split into two paths, one
where the bare qubit frequency is shifted by a fixed of which undergoes a π/2-radian phase rotation. To perform
amount g 2 /∆, known as the Lamb shift †† as well as an analog modulation, the two signals in each path are multiplied
amount proportional to the number of photons populat- at a mixer, yielding the outputs I(t) and Q(t), each having
ing the resonator52,215 . This effect is known as the ac- frequencies ωRO ±ωLO . I(t) and Q(t) are then low-pass-filtered
Stark shift. It has the consequence that photon number (time averaged) to yield IIF (t) and QIF (t) at the intermediate
frequency ωIF = |ωRO −ωLO |, and subsequently digitized using
fluctuations (noise) in the resonator induce small shifts
an analog-to-digital (ADC) converter. If ωIF 6= 0, then digital
of the qubit frequency, slightly bringing the qubit out of signals IIF [n] and QIF [n] are further digitally demodulated
its rotating frame and thus causing dephasing147 . This using digital signal processing (DSP) techniques to extract
means that spurious photon occupation and fluctuation the amplitude and phase of the readout signal.
in the resonator, be it thermal or coherent photons, shift
the qubit frequency and causing dephasing319,324 . For
this reason, it is important to make sure that the proces-
sor is properly thermalized107 , and its control lines well maximal when the resonator is probed just in-between
filtered325 and attenuated148 , to reduce photon number the two qubit-state dependent resonance frequencies163 ,
fluctuation. |0i |1i
ωRF = (ωr + ωr )/2. In this case, the reflected magni-
tude is identical for |0i and |1i, and all information about
the qubit state is encoded in the phase θ, see dashed
B. Measuring the resonator amplitude and phase
line in Fig. 19(b). In turn, the qubit-resonator detun-
ing should be designed to obey the criterion for maximal
In the previous section, we outlined the underlying state visibility, χ = κ/2, which is maximized for phase
physics behind the dispersive readout technique, in which measurements while constraining qubit dephasing.
we concluded that the qubit induces a state-dependent Once we have picked the resonator probe frequency, the
frequency shift of the resonator. We now focus our at- quantum dynamics of the qubit can be mapped onto the
tention on how to probe the resonator to “read out the phase of the classical microwave response. In the follow-
qubit,” that is, best distinguish the two classical res- ing, we discuss how we can use heterodyne detection to
onator signatures corresponding to our qubit states, see measure the phase of the resonator response. We assume
Fig. 19(b)-(c). that the reader is already somewhat familiar with basic
The readout circuit can be set up in measuring ei- mixer operations, such as modulation and de-modulation
ther reflection or transmission. The best state discrim- of signals. For interested readers, we refer to Ref. 326.
ination is obtained by maximizing the separation be-
tween the two states in the (I, Q)-plane, i.e. the in-
phase and quadrature component of the voltage, see
Fig. 19(c). It can be shown that this separation is 1. Representation of the readout signal

A readout event commences with a short microwave

tone directed to the resonator at the resonator probe fre-
∗∗ It has been demonstrated that it is still possible to read out the quency ωRO . After interacting with the resonator, the
qubit state by applying a very strong resonator drive tone, even-
reflected (or transmitted) microwave signal has the form
though this readout scheme is not QND321
†† It is worth mentioning that the observed qubit frequency is always

the Lamb-shifted frequency and not the bare qubit frequency. s(t) = ARO cos(ωRO t + θRO ), (148)
 44

where ωRO is the carrier frequency used to probe the res- operation is a combination of a balanced (50-50) beam-
onator. ARO and θRO are, respectively, the qubit-state- splitter followed by optical photodetectors, as shown in
dependent amplitude and phase that we want to measure. the inset of Fig. 22(a). The signal and local-oscillator op-
One can equivalently use a complex analytic representa- tical fields are first combined at the beamsplitter, yield-
tion of the signal, ing superpositions of both fields, and then detected at
n o the photodetectors, which act as square-law devices. To
s(t) = Re ARO ej(ωRO t+θRO ) , (149) build intuition for how this works, tbe square of the sum
of two electric fields (E1 + E2 )2 = E12 + E22 + 2E1 E2 has
= Re {ARO cos(ωRO t + θRO ) + j sin(ωRO t + θRO )}
a cross term that is the multiplication of the two fields.
where Re takes the real part of an expression, e.g., We refer the reader to Ref. 327 for further details.
Re[exp(jx)] = Re(cos x + j sin x) = cos x.
To gain intuition, we can rewrite Eq. (149) in a static
“phasor” notation that separates out the time depen- 3. Homodyne demodulation
dence exp(jωRO t),
  One direct means to extract I and Q is to perform a
  microwave homodyne measurement using an analog I-Q
s(t) = Re ARO ejθRO ejωRO t , (150) mixer of the type shown in Fig. 21. In an analog homo-
dyne measurement, the local oscillator (LO) is chosen to
| {z } 
phasor
be at the carrier frequency ωLO = ωRO . Upon mixing, I(t)
where the phasor ARO exp(jθRO ) ≡ ARO ∠θRO is a short- and Q(t) contain terms at both DC (ωIF = 0) and terms
hand that fully specifies a harmonic signal s(t) at a known at twice the carrier frequency. Time-averaging (filtering)
frequency ωRO . To perform qubit readout, we want to I(t) and Q(t) directly yield the DC terms IIF (t) = I and
measure the “in-phase” component I and a “quadrature” QIF (t) = Q:
component Q of the complex number represented by the
phasor, 1 T
Z
I= dt sI (t)yI (t)
ARO ejθRO = ARO cos θRO + jARO sin θRO (151) T 0
ARO ALO
≡ I + jQ (152) = cos(θRO ), (153)
8
to determine the amplitude ARO and the phase θRO . 1
Z T
Q= dt sQ (t)yQ (t)
T 0
ARO ALO
2. I-Q mixing = sin(θRO ), (154)
8
One direct means to extract I and Q is to perform a where T is a time interval taken to be an integer num-
homodyne or heterodyne measurement using an analog I- ber of periods of the readout signal. I and Q are then
Q mixer. Figure 21 shows a basic electrical schematic sampled and used to calculate the amplitude and phase:
of an I-Q mixer. The readout signal s(t) and a refer- p
ence local-oscillator signal y(t) = ALO cos ωLO t are fed ARO ∝ I 2 + Q2 , (155)
into the mixer via the RF and LO mixer ports. The θRO = arctan(Q/I). (156)
mixer then equally splits the signal and local oscillator
into two branches and multiplies them in the following Note that the global value of ARO or θRO is not what mat-
way: in the I-branch, the signal sI (t) = s(t)/2 is mul- ters; what matters is the change in ARO and θRO between
tiplied by the local oscillator yI (t) = (ALO /2) cos ωLO t; the qubit being in state 0i and state 1i. For example,
and in the Q-branch, the signal sQ (t) = s(t)/2 is multi- the value of A leaving the resonator and the value G × A
plied by a π/2-phase-shifted version of the local oscilla- reaching a measurement stage are different, where G rep-
tor, yQ (t) = −(ALO /2) sin ωLO t. The “-” sign arises from resents the net gain in the measurement amplifier chain.
the choice of using a A(cos ωt + φ) as the reconstructed However, the gain is the same, independent of the qubit
real signal. At the I and Q ports, the output signals I(t) (0)
state, whereas A may be different, e.g., ARO = G × A|0i
and Q(t) contain terms at the sum and difference frequen- (1)
or ARO = G × A|1i . Similarly, the propagation phase
cies, generally referred to as an intermediate frequency,
φ accumulated while a signal travels between the res-
ωIF = ωRO ± ωLO . The resulting signals are low-pass fil-
onator and the measurement stage is also independent
tered, passing only the terms at the difference frequency,
of the qubit state, and simply imparts a phase offset to
IIF (t) and QIF (t), which are then digitized. After digital (0)
signal processing, one obtains the static in-phase (I) and the qubit-induced phase shift, e.g., θRO = θ|0i + φ or
(1)
quadrature (Q) components, from which one calculates θRO = θ|1i + φ.
the amplitude ARO and the phase θRO . Homodyning works in principle, but there are two
Microwave mixers use square-law-type diodes to im- drawbacks. First, signals directly demodulated to DC
plement multiplication. The optical analog of a mixer may be subject to lower signal-to-noise ratios, since they
 45

(a) ANALOG DEMOD. (b) DATA SAMPLING (c) DIGITAL SIGNAL PROCESSING (DSP)

1 τrd τs

(a.u.)
ωIF
ωRO 0
Q

ωLO

-1
Integrate
1
FROM
CRYOSTAT
s(t) ωRO I ωIF (a.u.)
R Q 0
L
ωLO
-1
LO 106
Time, τ (a.u.)

FIG. 22. Schematic of the heterodyne detection technique. (a) The signal with frequency ωRF from the cryostat is mixed with
a carrier tone with frequency ωLO , yielding two quadratures at a down-converted intermediate frequency ωIF = |ωRO − ωLO |, and
90◦ out-of-phase with each other. (b) The two signals are passed into two different analog-to-digital converter (ADC) channels.
To avoid sampling the resonator transient, some readout delay (τrd ) corresponding to the resonator linewidth may be added,
and the two signals are sampled for a time τs . In this case, the white dots represent the sampled points. (c) The sampled
traces are post-processed and after some algebra, the sampled data points are averaged into a single point in the (I, Q)-plane.
To extract statistics of the readout performance, i.e. single-shot readout fidelity, a large number of (I, Q)-records are acquired,
yielding a 2D-histogram, with a Gaussian distributed spread given by the noise acting on the signal.

fight against 1/f electronics noise, as well as any other 4. Heterodyne demodulation
noise signals that may have inadvertently been demod-
ulated (e.g., via a square-law detector). The second is
that homodyning is not compatible with frequency di-
vision multiplexing (FDM), where a single pulse can be
used to interrogate N resonators at different frequencies
by applying tones at each resonator frequency using the
superposition principle, e.g.,

N
(i) (i) In a heterodyne scheme, a local oscillator at frequency
X
s(t) = ARO cos(ωRO t + θ(i) ). (157)
i=1
ωLO is offset by an intermediate frequency ωIF to target a
unique readout frequency ωRO . Up-conversion techniques
such as single-sideband modulation with suppressed car-
rier (SSB-SC) using balanced I-Q mixers (operated in re-
Homodyning an FDM signal will put all resonator signals verse compared with Fig. 21) are commonly used to cre-
at DC, and once downconverted, they cannot be differen- ate such readout signals. We refer the reader to Ref. 326
tiated. To work around this, it is generally advantageous for more information on how to create such pulses.
to use heterodyning, which uses a two-step demodulation Here, we want to extract ARO and θRO (or their scaled
process via an intermediate frequency ωIF . Such a scheme and offset versions) from the reflected/transmitted tone
is easily compatible with the concept of FDM, because a using a heterodyning scheme. The first step is to per-
readout signal is first demodulated to unique IF frequen- form analog I-Q mixing, as illustrated in Fig. 22(a). In
(i)
cies ωIF , and then digitally demodulated to extract each contrast to the homodyning case, here, the local os-
(i)
ARO and θ(i) . In the following, we will consider N = 1 cillator and readout tone are at different frequencies,
for simplicity, but the process is applicable to larger N ωIF = |ωRO − ωLO | > 0. Mixing the LO and RO sig-
provided the frequencies a sufficiently spaced to avoid nals yields the signals I(t) and Q(t) with terms at both
interference with one another during the demodulation sum and difference frequencies. Filtering out the sum fre-
process. quencies using low-pass filtering (time averaging) yields
 46

the IF signals: Fig. 22(c-d),

1 T
Z zIF [n] = IIF [n] + jQIF [n] ≡ VI [n] + jVQ [n] (164)
IIF (t) = (dt) sI (t)yI (t) ARO ALO
T 0 = [cos(ΩIF n + θRO ) + j sin(ΩIF n + θRO )]
ARO ALO 8
= cos(ωIF t + θRO ) (158) (165)
8
ARO ALO jθRO jΩIF n
1
Z T
= e e (166)
QIF (t) = (dt) sQ (t)yQ (t) 8
T 0
ARO ALO where the digital in-phase and quadrature signals are rep-
= sin(ωIF t + θRO ). (159) resented here as the voltages VI [n] and VQ [n] sampled
8
by the ADC, and we have separated the static phasor
As before, we have omitted any offset phases from the (ARO ALO /8) exp[jθRO ] from the rotating term exp[jΩIF n].
LO or from the wave propagation between the resonator One can digitally demodulate the time series zIF [n] by
and the measurement. Again, these offset values are not multiplying by the complex conjugate of the oscillatory
what matters; it is the change in ARO and θRO with a exponential,
change in qubit state that allows state discrimination.
The analog-demodulated IIF (t) and QIF (t) are now os- z[n] = zIF [n]. ∗ e−jΩIF n (167)
cillating at a frequency that is generally low enough to
where .∗ indicates a point-by-point multiplication, and
be digitized using commonly available analog-to-digital
the result is a vector of length M of nominally identical
converters (ADCs). The resulting digital signals are now
values of the phasor – one for each sample point – with a
written as IIF [n] and QIF [n],
small amount of additive noise due to noise in measure-
ment chain, digitization errors, etc. A singular phasor
ARO ALO
IIF [n] = cos(ΩIF n + θRO ) (160) value is then estimated by taking average,
8
ARO ALO 1 X
QIF [n] = sin(ΩIF n + θRO ), (161) z̄[n] = z[n] (168)
8 M
ARO ALO jθRO
where n = t/∆t indexes the sample number of the = e . (169)
8
continuous-time signals IIF (t) and QIF (t), ΩIF = ωIF ∆t is
the digital frequency, and ∆t is the sampling period (typ- Such “single-shot measurements” may then be repeated
ically around 1 ns). Pulsing the resonator is necessarily a large number of times to obtain an ensemble average
accompanied by a ring-up time, related to the quality hz̄[n]i.
factor of the resonator, and the first few samples may
decrease overall signal-to-noise. Consequently, a delayed
window of samples [n1 : n2 ] is often used to perform the C. Weak and strong qubit measurements: Impact of noise
second digital demodulation of the discrete-time signals
IIF [n1 : n2 ] and QIF [n1 : n2 ]. Note that more compli- In quantum measurements, noise plays an essential role
cated windowing functions may also be used to improve as it dictates the fidelity of its outcome128,328 , recall Fig.
state discrimination, but here we use a simple boxcar [see 22(c). In the absence of noise, any non-zero dispersive
Fig. 22(b)]. shift (resulting in a resonator field displacement) would
Digital demodulation comprises the point-by-point suffice to unambigously separate the qubit states, given
multiplication of IIF [n1 : n2 ] and QIF [n1 : n2 ] by cos ΩIF n a properly chosen resonator linewidth. In practice, how-
and sin ΩIF n. Averaging the resulting time series elimi- ever, the outcome of the quantum measurement is gener-
nates the 2ΩIF component while retaining the DC com- ally Gaussian distributed in the (I, Q)-plane due to pres-
ponent, as in a homodyne measurement, one obtains ence of classical and quantum noise. In this section, we
review the main sources of noise, as well as how it im-
n2
1 X ARO ALO pedes our ability to extract information from the quan-
I= IIF [n] cos[ΩIF n] = cos θRO , (162) tum system. For a rigorous discussion of noise and quan-
M n 16
1 tum measurements, the interested reader is referred e.g.
n2
1 X ARO ALO to the work by Clerk et al 128 and to the textbook by
Q= QIF [n] sin[ΩIF n] = sin θRO , (163) Haus127 .
M n 16
1 The total noise added to the signal has multiple origins.
One part of the noise is associated with the microwave
where M = n2 − n1 + 1. As before, I and Q can then be signal used to probe the resonator, where each photon
used to find ARO and θRO . has an intrinsic quantum noise power of ~ω/2 per unit
The same procedure may be view in the complex I − Q bandwidth. Another contribution comes from the phase-
plane by the analytic function zIF [n], as illustrated in preserving amplifiers, adding both classical noise and at
 47

(a) relative separation in the complex plane is maximized‡‡ .

Integrated det. signal (a.u.)
The line that is used to separate between |0i and |1i is
called a separatrix.
θ1 The noise can now be quantified by comparing the
widths of the Gaussian probability distribution sur-
rounding the mean with the peak separation in the
(I, Q)-plane, thus defining a signal-to-noise ratio SNR =
δθ/(∆θ1 + ∆θ0 ), see Fig. 23(a), with δθ = |θ1 − θ0 | repre-
θ0 senting the signal and ∆θ0 , ∆θ1 represent the noise (2σ)
of each distribution. The SNR allows us to distinguish
between a weak and a strong quantum measurement, as
illustrated in Fig. 23(b)-(d).
Sampling time, τs (a.u.) In a weak measurement, the probabilities are broadly
distributed as compared to their relative separation
(b) (c) (d) (SNR < 1), which means that only partial information
Integrated det. signal (a.u.)

of the quantum state is revealed to the observer, see

Fig.23(b). In a strong measurement, on the other hand,
the quantum state is collapsed onto one of the two eigen-
states. In this case, the outcome of the measurement can
be distinguished unambigously, which is reflected in two
fully separated distributions (SNR > 1), see Fig. 23(d).
In many applications of quantum measurements, it
is necessary to unambigously (and with high fidelity)
tell the outcome without repeating the readout measure-
Weak: Strong: ment. This is known as single-shot readout and it often
SNR < 1 SNR > 1 requires the use of a parametric amplifier – a preamplifier
Counts (a.u.) Counts (a.u.) Counts (a.u.) used to increase system SNR – which is further discussed
in Sec. V E 3.
FIG. 23. (a) Qubit state distribution throughout the course Assuming that the widths of the two distributions are
of sampling the readout signal, in the presence of noise. The identical, ∆θ0 = ∆θ1 = ∆θ, the separation error can
separation between the peaks (solid lines) increases
√ linearly in be calculated by deriving the weight of the overlapping
time, whereas the peak widths only increase as t. Image in- region of the Gaussian distributions as163
spired by Clerk et al 128 . The three black arrows represent line
cuts for three sampling times: (b) For short sampling time, ∞
1 (θ − θ1 )2
Z  
the states are not separated, resulting in a weak meaurement
sep = p exp − dθ
(SNR < 1). (c) After a longer sampling time, the peaks starts 2π(∆θ)2 θ=
θ0 −θ1
2
2(∆θ)2
to get separated, (d) finally getting fully resolved, resulting
in a strong measurement (SNR > 1). " #
1 |θ0 − θ1 |
= erfc p , (170)
2 2 2(∆θ)2

where erfc(x) denotes the complementary Gaussian error

least ~ω/2 of noise as required by Heisenberg’s uncer- function, defined as
tainty relation. Finally, any attenuation of the signal
prior to the first amplifier will appear as added noise.
∞
Combined, these noise sources amount to a system noise 2
Z
2
erfc(x) = 1 − √ e−t dt. (171)
temperature, which can be characterized using a sensitive π x
thermometer, such as a shot-noise tunnel junction329 or
a qubit330 as a sensor. Using the erfc in Eq. (171), the separation error in
Eq. (170) can be compactly expressed in terms of the
The noise results in time-dependent fluctuations of the signal-to-noise ratio,
measured signal, which in turn translates into uncer-
tainty in our demodulated signals, see Fig. 22(c). This
1 SNR
 
can be intuitively understood by considering that our sep = erfc (172)
heterodyne detection method requires us to sample for a 2 2
finite amount of time.
To quantify the impact of the noise on our measure-
ment, we first project the distributed (I, Q) data – cor- ‡‡ When analyzing the readout data, we have the freedom to choose
responding to |0i and |1i – onto the axis for which their projections.
 48

Note, however, that the separation error between the its environment. This is known as the Purcell effect331 ,
two state distributions only tells us the signal-to-noise and is an important consideration when designing qubit-
ratio of our detection scheme. On top of the separation resonator systems332 . The portion of spontaneous emis-
error, fidelity is reduced if the qubit relaxes (or is excited) sion that is mediated by the resonator describes how
during the readout. This will result in a count on the qubit relaxation is enhanced by the resonator Q when on-
“wrong” side of the threshold. This leads to an additional resonance, and suppressed off-resonance. The aim of this
constraint on the readout; The readout cycle needs to be section is twofold: first, we develop an intuition for how
completed on a timescale much shorter than the qubit the Purcell decay limits qubit coherence, and second, how
relaxation time. to properly mitigate this limitation by designing a so-
In summary, we see that to optimize the qubit read- called Purcell filter, which modifies the impedance seen
out fidelity, the readout needs to fulfill the following two by the qubit through the readout resonator. This allows
requirements: us to maintain fast readout, while protecting the qubit
from relaxing into its environment.
• Fast readout: The readout cycle needs to be com- If we would just choose qubit and resonator operation
pleted within a time that is short compared with frequencies guided by the resonator linewidth κ, qubit-
the qubit coherence time. The longer the readout resonator coupling g, and the amount of dispersive shift
time, the more likely the qubit is to relax, thus re- χ, we would reduce the detuning between the qubit and
ducing readout fidelity. the resonator, thus maximizing the dispersive shift (recall
• High signal-to-noise ratio: The signal-to-noise Fig. 20). However, this presents a trade-off between two
ratio needs to be sufficiently large to suppress the important system parameters; on one hand, we want the
state separation errors below an acceptable limit qubit to be isolated from the resonator environment off-
where it does not limit the readout fidelity. resonant to avoid Purcell-enhanced decay. On the other
hand, looking at the dispersive shift, we want the two
In sections V D and V E 3, we review how these two rates, g and κ to be strong, yielding larger dispersive
conditions are met by carefully engineering the signal shift as well as short resonator transient and thus a faster
path of the readout circuitry. readout.
Fortunately, when operating in the dispersive regime,
the qubit and resonator are far detuned from each other
D. “Purcell filters” for faster readout ∆  g, κ, which means that their impedance (environ-
ment) can be independently engineered through filter
To ensure high-fidelity readout performance, it is im- design. In essence, one designs a filter to have strong
portant to perform single-shot readout at a timescale coupling to the readout port at the resonator frequency
much shorter than the qubit coherence time, τro  T1 . (large κ), but isolates the qubit from its environment at
This motivates us to: (i) make the resonator linewidth the qubit frequency333,334 . In other words, an impedance
wide, thus reducing its ring-up time, τrd , and (ii) keep transformation.
the integration time τs as short as possible, see Fig.22(b). Depending on the design of the readout for the quan-
The ability to isolate a quantum system from decohering tum processor to which the filter should be coupled,
into its environment while, at the same time, being able there are different ways to design a Purcell filter; such as
to read out its state in a short time represents two con- quarter-wave stubs333 , low-Q bandpass filters66,334 , and
tradictory criteria, which must be traded-off328 . stepped-impedance filters335 . Which one is optimal de-
Even though dispersive readout (in the few-photon pends on system properties such as qubit-resonator de-
limit) has only a small back-action on the qubit state, tunings, required bandwidth, and allowed insertion loss.
the qubit will still suffer from T1 -relaxation while we are The most promising Purcell filter designs are the ones
performing a measurement. In fact, this “decay during that allow for frequency multiplexing, such as the low-Q
the readout” often limits the readout fidelity, reducing it bandpass filter design66,334 , which in addition to Purcell
to filtering has the function of a quantum bus, connecting
several frequency-multiplexed readout resonators sharing
the same amplifier chain.
F (τro ) = 1 − e−τro /T1 , (173)
The Purcell effect can be framed in terms of Fermi’s
where τro = τrd + τs /2 denotes the total time for the golden rule, where noise in the environment causes the
readout, consisting of the readout delay τrd due to the qubit to decay with some probability. We can gain intu-
resonator transient, and half the sampling time τs /2. The ition about the Purcell effect (as well as how the qubit can
fidelity drop in Eq. (173) can be interpreted as a man- be protected from it) by replacing the Josephson junction
ifestation of the competition between the time scales at in the qubit circuit with an ac-current source, outputting
which our quantum information reaches our detector or I(t) = I0 sin(ωt), with I0 = eω and study the rate at
the environment first. which power is lost into an environmental load resistor
The limitation of qubit coherence originates from an R = Z0 = 50 Ω, see Fig. 24(a).
enhanced spontaneous emission of photons, induced by Expressing the power lost in the resistor as P =
 49

I02 (Cg /CΣ )2 R = (eωβ)2 Z0 , with β = Cg /CΣ , the qubit We can now introduce the Purcell filter [Fig. 24(c-d)]
Purcell decay rate into the continuum can be written as in between the readout resonator and the 50 Ω environ-
ment, leading to a reduction of the decay rate according
to333
1 P (βeω)2 Z0 g2
Purcell
γenv = = = = . (174)
T1 ~ω ~ω ω
To protect the qubit from decaying into the 50 Ω envi-  2   
ronment (as well as for deploying our dispersive readout) g ωq ωr
Purcell
γres-filter-env =κ , (179)
we can now add a resonator in parallel with the qubit, ∆ ωr 2QF ∆
see Fig. 24(b). The presence of the resonator has the
effect of shaping the impedance at the qubit frequency,
which in turn modifies the decay rate in Eq. (174) into where QF denotes the quality factor of the Purcell filter.
This is schematically depicted in Fig. 24(d), where the
Purcell filter is placed around the resonator frequency,
g 2 Re[Zr (ω)] while far detuned from the qubit.
Purcell
γres-env = , (175)
ω Z0
where Zr (ω) denotes the impedance of the shunted res-
onator. We can express the real-part of the impedance (a) I(t) g (b) I(t) g κ
in terms of the resonator quality factor Q = ωr /κ and
Cg
qubit-resonator detuning ∆ = ωq − ωr ,
CΣ Z0 Z0
ωq ωq ωr
QZ0 Qubit Env. Qubit Res. Env.
Re[Zr (ω)] = . (176)
1 + 2(∆/κ)2
(c) I(t) g κ
Now, by substituting Eq. (176) into Eq. (175), we
see that the Purcell decay rate for the qubit depends on
CΣ Z0
the detuning between the resonator and the qubit. This
is intuitive, since the resonator can be thought of as a ωq ωr ωp
bandpass filter, with center frequency ωr and bandwidth Qubit Res. Purcell filter Env.
κ. For resonant condition, i.e. when ∆ = 0, the emission
rate into the resonator takes the form (d)
Purcell filter
g 2 Re[Zr ] g2 g2
Magnitude (a.u.)

Purcell
γres-env = = Q= . (177)
ωr Z0 ∆=0 ωr κ
In the dispersive regime ∆  g, κ, which is also rele-
vant for us in the context of qubit readout, we can make Qubit Resonator
the approximation Re[Zr ] ≈ QZ0 (κ/∆)2 , yielding the
familiar expression for the Purcell decay rate in circuit
QED332
ωq ωr
Frequency, (a.u.)
g 2 Re[Zr ] g2  κ 2  g 2
Purcell
γres-env = = Q = κ.
ωr Z0 ∆g,κ ωr ∆ ∆ FIG. 24. (a) Circuit representation of qubit (orange) cou-
(178) pled to an environment (blue) with a load resistor, Z0 , via a
The relation for the Purcell limit in Eq. (178) thus capacitor Cg . To study the decay rate, the Josephson junc-
provides us with a useful guide on how to design the cou- tion has been replaced with a current source, I(t). (b) By
pling rates g and κ, as well as how large qubit-resonator adding a resonator (red) with frequency ωr in-between the
detuning ∆ is necessary to avoid the Purcell limit. qubit and the 50 Ω environment, we get the case found in reg-
In recent years, however, the intrinsic coherence times ular dispersive readout. (c) A Purcell-filter (green) is added
for superconducting qubits have reached above 100 µs, to the circuit, providing protection for the qubit, while al-
lowing the resonator field to decay fast to the environment.
recall Sec. II, imposing practical limitations on how to
(d) Transmission spectrum of a Purcell filter (dashed green),
simultaneously optimize g and κ, to render fast readout centered around the resonator frequency (red arrow), whereas
without compromising the qubit coherence. Considering the qubit frequency (orange arrow) is far detuned.
the parameters in Eq. (178), it is not possible to just
increase the bound on the relaxation time T1 , without at
the same time trading off the readout speed and contrast.
 50

E. Improve signal-to-noise ratio: Parametric amplification fields and therefore considered to be coherent light com-
prising microwave photons. As such, they must obey the
In light of the aforementioned limited signal-to-noise commutation relations127,328,339,340
ratio associated with the low photon number of the dis-
persive qubit readout, and the short sampling time, the
[ain , a†in ] = [aout , a†out ] = 1, (182)
noise temperature of the amplifier chain plays a crucial
role in determining the fidelity of the measurement. from which it can be shown that it is not possible to
A useful benchmark for quantum measurements is the simultaneously amplify both quadratures of ain without
quantum efficiency, defined as also adding noise. This is known as Caves theorem after
the work by Caves328 , based on earlier work by Haus and
Mullen339 . This can be seen by considering the scattering
~ωRF
ηSQL = , 0 < ηSQL < 1, (180) relation between the input and output microwave fields
kB Tsys
√
which quantifies the photon energy to the system noise aout = Gain . (183)
temperature Tsys , thus yielding a measure of how close
the signal is to the standard quantum limit (SQL), as The gain relation in Eq. (183) constitutes our ideal
imposed by Heisenberg’s uncertainty relation, adding scenario for an amplifier process. However, the problem
1/2 photon of noise when ηSQL approaches unity. Since is that that this relation does not satisfy the commutation
the energy of each microwave photon is much smaller relation in Eq. (182). To satisfy this relation, we need to
than that of optical photons, it is not easy to build also take into account the vacuum fluctuations of another
a single-photon detector operating in the microwave mode128,341–343 – called the idler mode bin , also satisfying
domain336,337 . Instead, for heterodyne detection in cir- the same communtation relation [bin , b†in ] = 1. To satisfy
cuit QED, a set of cascaded microwave amplifiers are the commutation √ relation, the idler mode is amplified by
used. The system noise temperature for the amplifier the gain factor G − 1. For large gain, it can be shown
chain can be expressed in terms of the individual gain fig- that a minimum amount of half a photon of noise√~ω/2
ures Gn and noise temperatures TN,n of each constituent needs to be added to a signal amplified with gain G.
amplifier338 Finally, taking the idler mode into account, the scatter-
ing relation for the coherent output field takes the form
TN,2 TN,3 √ √
Tsys = TN,1 + + + ... (181)
G1 G1 G2 aout = Ga + G − 1b†in . (184)
| {z in} | {z }
Amplification Added idler noise
where n = 1, 2, 3, ... denotes the order of the amplifiers,
starting from the qubit chip. From Eq. (181), we see that Generally, this process results in a so-called phase-
the noise temperature Tsys is dominated by the noise con- insensitive parametric amplification process, in which
tribution from the first amplifier, whereas the gain of the both quadratures of the input field gets equally ampli-
first amplifier has the effect of suppressing the noise from fied. This is illustrated in Fig. 25, where the in-phase
the second amplifier, and so on. If the first amplifier is a (Iin ) and quadrature (Qin ) components of the fields are
low-noise high-electron mobility transistor (HEMT) am- plotted, before and after the parametric amplifier.
plifier (TN ≈ 2 K), the system noise temperature when Considering the amplification process in Eq. (184), we
implemented in a cryostat is around 7-10 K, correspond- can find a special case for the idler mode, for which noise-
ing to around 10-20 added photons of noise per signal less amplification can be accomplished for one of the two
photon around 5 GHz. In practice, this is generally too quadratures, but at the expense of adding more noise
much noise to perform single-shot readout. to the other, thus not violating Heisenberg’s uncertainty
This inherently poor signal-to-noise ratio has revived relation for the two field quadratures. This mode of op-
interest in developing quantum-limited parametric am- eration is known as phase-sensitive amplification, and is
plifiers (PA) – tailored for readout of superconducting obtained when the idler mode oscillates at the same fre-
qubits – featuring the ability to amplify small microwave quency as the signal (or a multiple thereof), but can be
signals, and adding only approximately the minimum shifted with an overall phase φ ∈ [0, 2π]. By substitut-
amount of noise allowed by quantum mechanics127,128,328 . ing the idler mode in Eq. (184) with bin = eiφ ain , the
scattering relation becomes

1. Quantum-limited amplification processes √ √

aout = Ga + e−iφ G − 1a†in . (185)
| {z in} | {z }
Amplification
In a linear, phase-insensitive amplifier, an input state
Phase-dep. noise

hain i is amplified to√an output state haout i, with an am- The overall phase factor allows us to tune the orienta-
plitude gain factor G. Microwaves are electromagnetic tion of the amplification (or de-amplification) by means
 51

(a) AMPLIFIER INPUT (b) MIXING PROCESSES (c) AMPLIFIER OUTPUT

Vac. noise

Pump
ωp

Signal
ωS

Idler
vac. ωi

FIG. 25. Schematic illustration of a quantum-limited, phase-preserving parametric amplification process of a coherent input
state, ain = Iin + iQin . (a) The state is centered at (hIin i, hQin i) and has a noise represented by the radii of the circles along the
real and imaginary axes, respectively. (b) Scattering representation of parametric mixing, where the signal and pump photons
are interacting via a purely √ dispersive nonlinear medium. (c) In the case of phase-preserving amplification, both quadratures
get amplified by a factor G, while (in the ideal case) half a photon of noise gets added to the output distribution (blue).
Image inspired by Flurin340 .

of the pump phase, thus allowing us to choose a quadra- exploited for qubit readout. Although many different
ture for which we want to reduce the noise, see Fig. 26. flavors of parametric systems exist, we here focus on the
Intuitively, this can be understood by considering the in- resonant implementations of the Josephson parametric
terference that occurs when two waves with the same fre- amplifier (JPA), serving as a good system for reviewing
quency are confined in space, where we obtain construc- the fundamental concepts around parametric amplifica-
tive or destructive interference, depending on the phase tion.
between the two waves. Due to this interference, the All parametric amplifiers operate based on one funda-
noise can be suppressed even below the standard quan- mental principle: the incoming signal photons are mixed
tum limit (without violating Heisenberg’s uncertainty re- with an applied pump tone via an intrinsic nonlinearity,
lation). This is known as single-mode squeezing and was by which energy from the pump is converted into signal
first observation in superconducting circuits by Yurke et photons and thereby providing gain. As we recall from
al.344 . In particular, after the theoretical prediction by Sec. II, such a nonlinearity can be engineered in the mi-
Gardiner345 , Murch et al. showed that the coherence crowave domain using Josephson junctions351 , and the
time of a qubit can be enhanced when the qubit is ex- resonant parametric amplifiers are built from slightly an-
posed to squeezed vacuum346,347 . Also two-mode squeez- harmonic oscillators.
ing was demonstrated by Eichler et al.348 , where the de- The first Josephson parametric amplifiers were built
modulation setup squeezes both quadratures of the ac- from a coplanar waveguide resonator, made nonlinear by
quired signal108 . adding a nonlinear Josephson contribution to its total
In the context of qubit readout, however, phase- inductance, see Fig. 27(a). The word parametric refers
sensitive amplification tends to be experimentally in- to the process of modulating (or pumping) one of the
convenient. This is mainly due to its phase-dependent parameters of the system’s equation-of-motion (such as
gain, which imposes stringent requirements on continu- frequency or damping) in time350,352,353 . The natural
ous phase-calibration of the readout signal. way to perform this parametric pumping is to modulate
For a detailed theoretical framework developed for the nonlinear Josephson inductance, which in turn has
quantum limited amplification, the reader is referred to thepeffect of modulating the resonator frequency ωr (t) =
earlier work by Roy and Devoret349 , Clerk et al.128 , and 1/ L(t)C.
Wustmann and Shumeiko350 . Depending on how the pumping is implemented, there
are two different mixing processes that can be exploited
in Josephson parametric amplifiers, which determines
2. Operation of Josephson parametric amplifiers the characteristics of the amplifier. These are illus-
trated in Fig. 27(b)-(c) and are referred to as current-
In this section, we review the basic operation charac- pumping 352,354–357 and flux-pumping 97,350,353,358–363 , re-
teristics of parametric amplifiers, and in particular the spectively. The type of mixing process that takes place
Josephson parametric amplifiers (JPAs), that have been depends on the leading order of the nonlinearity of the
 52

(a) (a) Josephson parametric amplifier (JPA)

Resonator SQUID
(b) (b) Current-pumping (4-wave mixing): 2ωp = ωs + ωi
30
ωp ωp
Voltage gain, |G| (dB)

G|ain|

Input + Pump
20

10
|ain|
0

-10 ωs ωr ωs ωr ωi
-20 (c) Flux-pumping (3-wave mixing): ωp = ωs + ωi

-30 |ain| ωp

Input
G|ain|
/2 3 /2 5 /2 7 /2 9 /2
Pump-phase angle, θ (rad)
ωp

Pump
FIG. 26. Phase-sensitive parametric amplification. (a) In
contrast to the phase-insensitive operation, phase-sensitive ωs ωr 2ωr ω s ω r ωi 2ωr
parametric amplification allows us to suppress the noise along
one axis. Consequently, the noise is added to the other
quadrature. (b) Voltage gain as a function of pump-phase FIG. 27. Circuit schematics and pump schemes of a Joseph-
angle, in which the amplification depends on the phase of the son parametric amplifier. (a) The device consists of a quarter-
pump, providing either amplification or de-amplification of wavelength resonator (blue), represented as lumped elements,
the quadrature voltage. shorted to ground via a Kerr-nonlinearity consisting of two
parallel Josephson junctions (orange) forming a SQUID. The
pump (red) can be applied in two ways; (b) either by mod-
ulating the current through the junctions (four-wave mixing)
system, as reflected in its Hamiltonian. In the follow- at the resonant frequency, ωp ≈ ωr , or (c) by modulating the
ing, we briefly review the difference between these two ac-flux Φac around a static dc-flux point Φdc using a separate
pump-schemes. fast-flux line (three-wave mixing). The flux pump is applied
In the current-pumped case, the dynamics of the sys- at twice the resonant frequency, ωp ≈ 2ωr .
tem has characteristics of a Duffing oscillator364 , with a
fourth-order nonlinear term in addition to the harmonic
oscillator term in its Hamiltonian
flux Φac through a SQUID loop, thereby modulating the
frequency of the resonator. This results in a three-wave
H = ωr c† c + Kc† c† cc, (186) mixing process, comprising three photons: one signal,
one idler, and one pump photon, with ωs + ωi = ωp , see
where c denotes the resonator field operator and K is the Fig. 27(c). Therefore, we see that the pump frequency is
“Kerr-nonlinearity”. This process is a so-called four-wave about twice that of the signal ωp ≈ 2ωs for ωs ≈ ωi . For
mixing process, since it mixes four photons: one signal degenerate, flux-pumped systems, the leading nonlinear-
(ωs ), one idler (ωi ), and two pump photons (ωp ), obeying ity is a third-order term, yielding a Hamiltonian
the energy conservation relation ωs + ωi = 2ωp , see Fig.
27(b). Pioneered by Yurke352 , this was the first demon-
stration of microwave amplification using a Josephson H = ωr c† c + K pc† c† + p† cc , (187)


parametric amplifier. When the signal and idler modes
are at the same frequency, the amplification is said to be where the p operator denotes the flux-pump mode. This
degenerate. This pumping scheme is the foundation for approach to building parametric amplifiers was devel-
the Josephson Bifurcation Amplifier (JBA), developed oped by Yamamoto et al.359 , as well as by Sandberg et
by Siddiqi et al.356,365,366 , which has been used to per- al.97 .
form single-shot qubit readout, by mapping the quantum The flux-pumping scheme has several practical advan-
states onto the high and low resonator field originating tages. First, the large detuning of the pump makes it eas-
from the sharp bifurcation point of the amplifier367 . ier to filter, isolating the readout signal as its passing into
In the other case, when the system is flux-pumped, the digitizer downstream and preventing the saturation
the parametric process is driven by threading a magnetic of following amplifier stages. Second, if the resonator is
 53

a quarter-wavelength resonator, it has no resonant mode distributing the nonlinearity across an array consisting of
at the pump frequency ωp , reducing spurious population many identical junctions, reducing the Kerr-nonlinearity
or saturation of the system as well as backaction on the by a factor 1/N 2 with N representing the number of junc-
qubits in the processor. Third, since the flux pump line tions in the array. This has been demonstrated by using
is a separate on-chip microwave line, no additional direc- a an array of SQUIDs in a resonator, rather than a single
tional coupler is needed. one360 .
Due to its rich dynamics, flux-pumping has also
proven a useful platform to study the quantum dy- However, despite the above mentioned engineering ef-
namics of Josephson parametric oscillators, both in the forts to improve the resonator-based JPAs, the most
context of qubit readout368–371 , the dynamical Casimir prominent approach to date is to get rid of the resonator
effect372–374 , and to better understand their complex altogether and, instead, construct a microwave analog to
nonlinear dynamics361,364,375–379 . optical parametric amplifiers, where kilometers of weakly
In addition to the degenerate parametric interactions nonlinear fibers are used. Such device is called a traveling
described above, parametric gain can be obtained be- wave parametric amplifier (TWPA) and was developed
tween different resonant modes; either between different to surmount the bandwidth and dynamic range limita-
modes of the same resonator329,380 , or in-between differ- tions of the resonator-based JPAs.
ent resonators381 , as with the Josephson parametric con- Although operated in similar way, the nonlinearity of
verter (JPC)340,382–386 . In addition to the possibility of TWPAs can be realized in different ways, such as the ki-
isolating and amplifying certain frequencies, the JPC can netic inductance of a superconducting film391–393 or using
implement frequency conversion for which it has some an array of Josephson junctions330,394,395 , through which
other areas of applications compared with other types of the four-wave mixing process is distributed across a non-
parametric amplifiers. linear lumped element transmission line, see Fig. 28(a).
The Josephson TWPA consists of a few thousand iden-
3. The traveling wave parametric amplifier tical unit cells, each comprising a shunt capacitor to
ground and a nonlinear Josephson inductor,p together
In the previously described JPA, parametric ampli- yielding a characteristic impedance of Z0 = LJ /C ≈
fication is realized using resonators that enhance the 50 Ω, see Fig. 28(a).
parametric interaction between the input signal and the The fact that the nonlinearity is distributed allows for
Josephson junction nonlinearity. Essentially, the Q- high saturation power, since each Josephson junction is
enhancement of the resonator forces each photon to pass accessed once. However, even though energy conserva-
through the junction on average Q times before leaving tion is satisfied, the four-wave mixing process in the de-
the resonator, thereby enhancing the non-linear interac- vice, there is a problem with phase (or momentum) con-
tion. Albeit proven to be able to reach near the standard- servation. This is associated with the system nonlinear-
quantum limit of noise for readout of a small number of ity as well as the large frequency detuning between signal
qubits, the future direction of the community is heading and pump photons, yielding a difference in phase-velocity
towards amplifier technologies which are compatible with between the two, which in turn gives rise to a non-flat
multiplexed readout of several qubits coupled to the same gain profile, as well as an overall reduction in gain394 .
amplifier chain65,66,387–389 . In this context, resonator-
based parametric amplifiers suffer from two major draw- Again, by taking inspiration from the dispersive en-
backs: First, the amplifier bandwidth is limited to the gineering developed in quantum optics and photonics,
resonator linewidth, typically ≈ 10 − 50 MHz, practically where the refractive index can be periodically altered to
limiting the number of multiplexed frequencies that can engineer the momentum of a transferred signal, the so-
be amplified. Second, since the Josephson nonlinearity lution to this phase-mismatch problem was introduced
is realized by a small number of junctions, the satura- by O’Brien et al.394 . By introducing resonators at peri-
tion power is low due to the interplay of higher order odic intervals of TWPA unit cells, the pump tone can be
nonlinearities, effectively taking the system outside its given a “momentum kick”, effectively slowing it down and
desired operation regime364,375,376,386 . In practice, this phase-matching the device by means of its wave vector.
limits how many readout resonators that can be simul- This technique is called resonant-phase matching (RPM),
tanously read out. see Fig. 28(d), and requires that the pump frequency
These two bottlenecks can, to a degree, be overcome is set on the left side of the dispersion feature (where
with microwave engineering. For instance, the linewidth the wave vector diverges), defined by the resonant fre-
can be made an order of magnitude wider by altering the quency of the phase-matching resonators. Note, finally,
impedance along the resonator. This is called a stepped- that broadband parametric amplification with high dy-
impedance transformer, where the impedance is ramped namic range has been demonstrated in other Josephson-
down from a matched 50 Ω at the capacitor down to a based circuits, e.g. the superconducting nonlinear asym-
small impedance at the SQUID390 shorting the device to metric inductive element (SNAIL) parametric amplifier
ground. Also the saturation power can be increased by (SPA)396 .
 54

(a) Josephson traveling wave parametric amplifier (JTWPA) ments with high signal-to-noise ratio. Putting these ad-
vances together, we hope that it is clear that the planar
RPM Res.
superconducting qubit modality is a promising platform
LJ CC for realizing near-term medium scale quantum proces-
IN C0 Lr
OUT sors. While we have focused on highlighting the advances
Cr made within the fields of realizing, controlling and read-
Z0 ing out planar superconducting qubits specifically used
for quantum information processing, there has of course
(b) also been tremendous activity in the surrounding fields.
ωs In this final section, we briefly mention a few of those
fields, and invite the reader to look into the references,
for further details.
ωp
Quantum annealing: Superconducting qubits
(c) also form the basis for certain quantum annealing
20
platforms397,398 . Quantum annealing operates by find-
TWPA
ing the ground state of a given Hamiltonian (typically
15 RPM a classical Ising Hamiltonian), and this state will
Gain (dB)

TWPA
correspond to the solution of an optimization problem.
10
By utilizing a flux-qubit type design (see Sec.II, the
5 company D-Wave have demonstrated quantum anneal-
ing processors88 which have now reached beyond 2000
0 qubits399 . The benchmarking of quantum annealers
0 2 4 6 8 10
Frequency (GHz) and attempts to demonstrate a quantum speedup for a
(d)
0.20
general class of problems is a highly active research field,
and we refer the reader, for example, to recent papers
Wavevector, k (a-1)

0.15 0.10
Ip=0.7I0 Refs. 400–402 and references therein.
Ip=0.5I0
Ip=0.3I0 Cavity based QIP: A parallel effort to the planar su-
0.10 0.08 perconducting qubits discussed in this review is the de-
5.85 6.00 6.10 velopment of 3D cavity-based superconducting qubits.
TWPA
0.05
RPM In these systems, quantum information is encoded
TWPA in superpositions of coherent photonic modes of the
0.00 cavity101 . The cat states can be highly coherent due
0 2 4 6 8 10
Frequency (GHz) to the inherently high quality factors associated with 3D
cavities102,403,404 . This approach has a fairly small hard-
ware overhead to encode a logical qubit405 , and lends
FIG. 28. (a) Simplified circuit representation of a Joseph-
son traveling wave parametric amplifier (JTWPA). The char- itself to certain implementations of asymmetric error-
acteristic impedance for each unit cell is set by the in-line correcting codes due to the fact that errors due to single-
Josephson inductor, LJ (orange) and the shunt capacitor, C photon loss in the cavity is a tractable observable to
(blue). A resonant LC-circuit (red) is used to phase match decode. Using this architecture, several important ad-
the four-wave amplification process. (b) Schematic of how the vances were recently demonstrated including extending
signal gets amplified in each unit cell as it propagates through the lifetime of an error-corrected qubit beyond its con-
the device. (c) Gain vs. frequency for a JTWPA, with and stituent parts100 , randomized benchmarking of logical
without the resonant phase matching (RPM). (d) Dispersion operations405 , a CNOT gate between two logical qubits406
relation of the TWPA, where the LC-resonators collectively as well as Ramsey interference of an encoded quantum
open up a stopband at the resonant frequency. By applying
error corrected qubit407 .
the pump close to this frequency, the wave vector of the pump
can be set to obtain a phase-matching. The optimal pump Cryogenics and software development: We briefly
frequency depends on the pump power, as indicated in the mentioned the electrical engineering, software develop-
inset. Image courtesy of Kevin O’Brien330,394 ment, and cryogenic considerations associated with the
control wiring and on-chip layout of medium-scale quan-
tum processors. While dilution refrigerators are now
readily available, off-the-shelf commercial products, the
details of how to optimally do signal-routing and rapid
VI. SUMMARY AND OUTLOOK
data processing in a scalable fashion, is also a field in
rapid development. However, with the recent demon-
In this review, we have discussed the phenomenal strations of enabling technologies such as 3D integra-
progress over the last decade in the engineering of su- tion, packages for multi-layered devices and supercon-
perconducting devices, the development of high-fidelity ducting interconnects408–414 , some of the immediate con-
gate-operations, and quantum non-demolition measure- cerns for how to scale the number of qubits in the su-
 55

perconducting modality, have been addressed. On the of remote entanglement, enabling quantum information
control software side, there currently exist multiple com- to be distributed across different nodes of a quantum pro-
mercial and free software packages for interfacing with cessing network429,430 .
quantum hardware, such as QCoDeS415 , the related py- Quantum computational supremacy: Finally, we men-
CQED416 , qKIT417 and Labber418 . However, many lab- tion one of the grand challenges for superconducting
oratories use software platforms developed in-house, of- qubits in the coming years: the demonstration of quan-
ten due to the concurrent development of custom-built, tum computational supremacy431 . The basic idea is to
highly specialized electronics and FPGA circuits (many demonstrate a calculation, using qubits and algorithmic
of these developments are not always published, but read- gates, which is outside the scope of classical comput-
ers may consult Refs. 190, 191, and 419 for three exam- ers (assuming some plausible computational complexity
ples). There is currently also a large ongoing develop- conjectures). For a recent review article, the reader is
ment of quantum circuit simulation and compiling soft- referred to Ref.432 . A first step towards an approach
ware packages. Packages such as Qiskit420 , Forest (with to demonstrating quantum supremacy was recently re-
pyQUIL421 ), ProjectQ422 , Cirq423 , OpenFermion424 , the ported, using 9 tunable transmons270 . It is expected
Microsoft Quantum Development kit425 provide higher- that with somewhere between 50-100 qubits433 , an ex-
level programming languages to compile and/or optimize tension of the protocol from Refs.270 and 434, will al-
quantum algorithms. For a recent review and compari- low researchers to sample from a classically intractable
son of these different software suites, we refer to Ref.426 distribution, and thereby demonstrate quantum compu-
and Ref.183 for a general review on advances in design- tational supremacy. The success of this program would
ing quantum software. Since the connectivity and gate constitute a phenomenal result for all of quantum com-
set of quantum processors can differ, details of the gate puting.
compilation implementation is an important non-trivial
problem for larger-scale processors. We note that some
of these software packages already interface directly with ACKNOWLEDGMENTS
quantum processors that are available online, supplied,
for example, via Rigetti Computing or the IBM Quan-
tum Experience. The authors gratefully acknowledge Mollie Kimchi-
Schwartz, Jochen Braumüller, Niels-Jakob Søe Loft, and
Quantum error correction: While the qubit lifetimes David DiVincenzo for careful reading of the manuscript
and gate fidelity have improved dramatically in the last and Youngkyu Sung for use of his time-dependent qubit
decades, there remains a need for error correction to drive simulation suite and useful feedback from the en-
reach large-scale processors. While certain strategies ex- tire Engineering Quantum Systems group at MIT. The
ist to extend the computational reach of current state-of- authors also acknowledges fruitful discussion with Anton
the-art physical qubits427 , for truly large-scale algorithms Frisk Kockum, Anita Fadavi Roudsari, Daryoush Shiri,
addressing practical problems, the quantum data will and Christian Križan.
have to be embedded in an error-correcting scheme. As This research was funded in part by the U.S. Army Re-
briefly mentioned in Sections IV F 2 and IV G 2, certain search Office Grant No. W911NF-14-1-0682; and by the
components of the surface code quantum error correcting National Science Foundation Grant No. PHY-1720311.
scheme have already been demonstrated in superconduct- P.K. acknowledges partial support by the Wallenberg
ing qubits (see e.g. Refs.66, 258, and 285). However, Centre for Quantum Technology (WACQT) funded by
the demonstration of a logical qubit with greater life- Knut and Alice Wallenberg Foundation. M.K. gratefully
time than the underlying physical qubits, remains an out- acknowledges support from the Carlsberg Foundation.
standing challenge. While the surface code is a promis- The views and conclusions contained herein are those of
ing quantum error correcting code due to its relatively the authors and should not be interpreted as necessarily
lenient fault tolerance threshold, it cannot implement a representing the official policies or endorsements of the
universal gate set in a fault-tolerant manner. This means US Government.
that the error-corrected gates in the surface code need to
be supplemented, for example, with a T gate, to become 1 R. Feynman, “Simulating physics with computers,” Int. J Theor.
universal. Such gates can be implemented by a tech- Phys 21, 467–488 (1982).
2 S. Lloyd, “Universal quantum simulators,” Science 273, 1073–
nique known as magic state distillation 428 . The process
of gate-teleportation, a pre-cursor to magic state distilla- 1078 (1996).
3 D. P. DiVincenzo, “The physical implementation of quantum
tion, has already been demonstrated using FPGA-based computation,” Fortschritte der Physik 48, 771–783 (2000).
classical feedback with planar superconducting qubits191 , 4 J. P. Dowling and G. J. Milburn, “Quantum technology: the
but showing distillation and injection into a surface code second quantum revolution,” Philosophical Transactions of the
logical state remains an open challenge. The develop- Royal Society of London A: Mathematical, Physical and Engi-
ment of new quantum codes is also a field in rapid devel- neering Sciences 361, 1655–1674 (2003).
5 T. D. Ladd, F. Jelezko, R. Laflamme, Y. Nakamura, C. Mon-
opment, and the reader may consult a recent review for roe, and J. L. O’Brien, “Quantum computers,” Nature 464, 45
more details e.g. Ref.184. Another important step to- (2010).
wards large-scale quantum processor architecture is that 6 M. A. Nielsen and I. L. Chuang, Quantum Computation and
 56

Quantum Information: 10th Anniversary Edition, 10th ed. “Quantum register based on individual electronic and nuclear
(Cambridge University Press, New York, NY, USA, 2011). spin qubits in diamond,” Science 316, 1312–1316 (2007).
7 C. Monroe and J. Kim, “Scaling the ion trap quantum proces- 30 E. Knill, R. Laflamme, and G. J. Milburn, “A scheme for effi-

sor,” Science 339, 1164–1169 (2013). cient quantum computation with linear optics,” Nature 409, 46
8 H. Bernien, S. Schwartz, A. Keesling, H. Levine, A. Omran, (2001).
H. Pichler, S. Choi, A. S. Zibrov, M. Endres, M. Greiner, 31 T. B. Pittman, B. C. Jacobs, and J. D. Franson, “Probabilis-

V. Vuletić, and M. D. Lukin, “Probing many-body dynamics tic quantum logic operations using polarizing beam splitters,”
on a 51-atom quantum simulator,” Nature 551, 579 (2017). Phys. Rev. A 64, 062311 (2001).
9 D. Loss and D. P. DiVincenzo, “Quantum computation with 32 J. D. Franson, M. M. Donegan, M. J. Fitch, B. C. Jacobs, and

quantum dots,” Phys. Rev. A 57, 120–126 (1998). T. B. Pittman, “High-fidelity quantum logic operations using
10 B. E. Kane, “A silicon-based nuclear spin quantum computer,” linear optical elements,” Phys. Rev. Lett. 89, 137901 (2002).
Nature 393, 133 (1998). 33 T. B. Pittman, M. J. Fitch, B. C. Jacobs, and J. D. Franson,
11 R. Vrijen, E. Yablonovitch, K. Wang, H. W. Jiang, A. Balandin, “Experimental controlled-not logic gate for single photons in the
V. Roychowdhury, T. Mor, and D. DiVincenzo, “Electron- coincidence basis,” Phys. Rev. A 68, 032316 (2003).
spin-resonance transistors for quantum computing in silicon- 34 M. H. devoret and J. M. Martinis, “Implementing qubits

germanium heterostructures,” Phys. Rev. A 62, 012306 (2000). with superconducting integrated circuits,” Quantum Informa-
12 R. de Sousa, J. D. Delgado, and S. Das Sarma, “Silicon quantum tion Processing 3, 163–203 (2004).
computation based on magnetic dipolar coupling,” Phys. Rev. 35 J. Q. You and F. Nori, “Superconducting circuits and quantum

A 70, 052304 (2004). information,” Physics Today 58, 42–47 (2005).

13 L. C. L. Hollenberg, A. D. Greentree, A. G. Fowler, and C. J. 36 R. J. Schoelkopf and S. M. Girvin, “Wiring up quantum sys-

Wellard, “Two-dimensional architectures for donor-based quan- tems,” Nature 451, 664 (2008).
tum computing,” Phys. Rev. B 74, 045311 (2006). 37 J. Clarke and F. K. Wilhelm, “Superconducting quantum bits.”
14 A. Morello, J. J. Pla, F. A. Zwanenburg, K. W. Chan, K. Y. Nature 453, 1031–42 (2008).
Tan, H. Huebl, M. Möttönen, C. D. Nugroho, C. Yang, J. A. 38 S. M. Girvin, Circuit QED : Superconducting Qubits Coupled to

van Donkelaar, A. D. C. Alves, D. N. Jamieson, C. C. Escott, Microwave Photons (Oxford University Press, Oxford, England,
L. C. L. Hollenberg, R. G. Clark, and A. S. Dzurak, “Single- 2009).
shot readout of an electron spin in silicon,” Nature 467, 687 39 J. Q. You and F. Nori, “Atomic physics and quantum optics

(2010). using superconducting circuits,” Nature 474, 589 (2011).

15 A. Imamoglu, D. D. Awschalom, G. Burkard, D. P. DiVincenzo, 40 W. D. Oliver and P. B. Welander, “Materials in superconducting

D. Loss, M. Sherwin, and A. Small, “Quantum information quantum bits,” MRS Bulletin 38, 816–825 (2013).
processing using quantum dot spins and cavity qed,” Phys. Rev. 41 J. M. Gambetta, J. M. Chow, and M. Steffen, “Building logical

Lett. 83, 4204–4207 (1999). qubits in a superconducting quantum computing system,” npj
16 J. R. Petta, A. C. Johnson, J. M. Taylor, E. A. Laird, A. Ya- Quantum Information 3, 2 (2017).
coby, M. D. Lukin, C. M. Marcus, M. P. Hanson, and A. C. 42 G. Wendin, “Quantum information processing with supercon-

Gossard, “Coherent manipulation of coupled electron spins in ducting circuits: a review,” Reports on Progress in Physics 80,
semiconductor quantum dots,” Science 309, 2180–2184 (2005). 106001 (2017).
17 D. Englund, D. Fattal, E. Waks, G. Solomon, B. Zhang, 43 X. Gu, A. F. Kockum, A. Miranowicz, Y. x Liu, and F. Nori,

T. Nakaoka, Y. Arakawa, Y. Yamamoto, and J. Vučković, “Microwave photonics with superconducting quantum circuits,”
“Controlling the spontaneous emission rate of single quantum Physics Reports 718-719, 1 – 102 (2017).
dots in a two-dimensional photonic crystal,” Phys. Rev. Lett. 44 M. H. Devoret, “Quantum Fluctuations,” in Les Houches Ses-

95, 013904 (2005). sion LXIII (Oxford University Press, 1997).

18 R. Hanson, L. P. Kouwenhoven, J. R. Petta, S. Tarucha, and 45 R. K. Naik, N. Leung, S. Chakram, P. Groszkowski, Y. Lu,

L. M. K. Vandersypen, “Spins in few-electron quantum dots,” N. Earnest, D. C. McKay, J. Koch, and D. I. Schuster, “Ran-
Rev. Mod. Phys. 79, 1217–1265 (2007). dom access quantum information processors using multimode
19 J. I. Cirac and P. Zoller, “Quantum computations with cold circuit quantum electrodynamics,” Nature Communications 8,
trapped ions,” Phys. Rev. Lett. 74, 4091–4094 (1995). 1904 (2017).
20 D. Leibfried, R. Blatt, C. Monroe, and D. Wineland, “Quantum 46 B. D. Josephson, “Possible new effects in superconductive tun-

dynamics of single trapped ions,” Rev. Mod. Phys. 75, 281–324 nelling,” Physics Letters 1, 251 – 253 (1962).
(2003). 47 B. D. Josephson, “Coupled superconductors,” Rev. Mod. Phys.
21 R. Blatt and D. Wineland, “Entangled states of trapped atomic 36, 216–220 (1964).
ions,” Nature 453, 1008 (2008). 48 Y. Nakamura, Y. A. Pashkin, and J. S. Tsai, “Coherent control
22 H. Häffner, C. Roos, and R. Blatt, “Quantum computing with of macroscopic quantum states in a single-Cooper-pair box,”
trapped ions,” Physics Reports 469, 155 – 203 (2008). Nature 398, 786 (1999).
23 R. Blatt and C. F. Roos, “Quantum simulations with trapped 49 D. Vion, A. Aassime, A. Cottet, P. Joyez, H. Pothier, C. Urbina,

ions,” Nature Physics 8, 277 (2012). D. Esteve, and M. H. Devoret, “Manipulating the quantum
24 D. Jaksch and P. Zoller, “The cold atom hubbard toolbox,” An- state of an electrical circuit,” Science 296, 886–889 (2002).
nals of Physics 315, 52 – 79 (2005), special Issue. 50 J. Q. You and F. Nori, “Quantum information processing with
25 M. Lewenstein, A. Sanpera, V. Ahufinger, B. Damski, superconducting qubits in a microwave field,” Phys. Rev. B 68,
A. Sen(De), and U. Sen, “Ultracold atomic gases in optical 064509 (2003).
lattices: mimicking condensed matter physics and beyond,” Ad- 51 T. Duty, D. Gunnarsson, K. Bladh, and P. Delsing, “Coherent

vances in Physics 56, 243–379 (2007). dynamics of a josephson charge qubit,” Phys. Rev. B 69, 140503
26 I. Bloch, J. Dalibard, and W. Zwerger, “Many-body physics (2004).
with ultracold gases,” Rev. Mod. Phys. 80, 885–964 (2008). 52 J. Koch, T. M. Yu, J. Gambetta, A. A. Houck, D. I. Schuster,
27 C. Gross and I. Bloch, “Quantum simulations with ultracold J. Majer, A. Blais, M. H. Devoret, S. M. Girvin, and R. J.
atoms in optical lattices,” Science 357, 995–1001 (2017). Schoelkopf, “Charge-insensitive qubit design derived from the
28 R. Hanson, O. Gywat, and D. D. Awschalom, “Room- cooper pair box,” Phys. Rev. A 76, 042319 (2007).
temperature manipulation and decoherence of a single spin in 53 M. J. Peterer, S. J. Bader, X. Jin, F. Yan, A. Kamal, T. J.

diamond,” Phys. Rev. B 74, 161203 (2006). Gudmundsen, P. J. Leek, T. P. Orlando, W. D. Oliver, and
29 M. V. G. Dutt, L. Childress, L. Jiang, E. Togan, J. Maze, S. Gustavsson, “Coherence and decay of higher energy levels
F. Jelezko, A. S. Zibrov, P. R. Hemmer, and M. D. Lukin, of a superconducting transmon qubit,” Phys. Rev. Lett. 114,
 57

010501 (2015). B. Bloom, A. Bradley, C. Bui, S. Caldwell, L. Capelluto,

54 L. DiCarlo, M. D. Reed, L. Sun, B. R. Johnson, J. M. Chow, R. Chilcott, J. Cordova, G. Crossman, M. Curtis, S. Desh-
J. M. Gambetta, L. Frunzio, S. M. Girvin, M. H. Devoret, and pande, T. El Bouayadi, D. Girshovich, S. Hong, A. Hudson,
R. J. Schoelkopf, “Preparation and measurement of three-qubit P. Karalekas, K. Kuang, M. Lenihan, R. Manenti, T. Manning,
entanglement in a superconducting circuit,” Nature 467, 574 J. Marshall, Y. Mohan, W. O’Brien, J. Otterbach, A. Papa-
(2010). george, J.-P. Paquette, M. Pelstring, A. Polloreno, V. Rawat,
55 H. Paik, D. I. Schuster, L. S. Bishop, G. Kirchmair, G. Cate- C. A. Ryan, R. Renzas, N. Rubin, D. Russel, M. Rust, D. Scara-
lani, A. P. Sears, B. R. Johnson, M. J. Reagor, L. Frunzio, L. I. belli, M. Selvanayagam, R. Sinclair, R. Smith, M. Suska, T.-
Glazman, S. M. Girvin, M. H. Devoret, and R. J. Schoelkopf, W. To, M. Vahidpour, N. Vodrahalli, T. Whyland, K. Yadav,
“Observation of high coherence in josephson junction qubits W. Zeng, and C. T. Rigetti, “Demonstration of universal para-
measured in a three-dimensional circuit qed architecture,” Phys. metric entangling gates on a multi-qubit lattice,” Science Ad-
Rev. Lett. 107, 240501 (2011). vances 4 (2018), 10.1126/sciadv.aao3603.
56 A. Shnirman, G. Schön, I. Martin, and Y. Makhlin, “Low- and 68 M. Tinkham, Introduction to Superconductivity (Dover publica-

high-frequency noise from coherent two-level systems,” Phys. tions, 2004).

Rev. Lett. 94, 127002 (2005). 69 M. D. Hutchings, J. B. Hertzberg, Y. Liu, N. T. Bronn, G. A.
57 J. Gao, J. Zmuidzinas, B. A. Mazin, H. G. LeDuc, and Keefe, M. Brink, J. M. Chow, and B. L. T. Plourde, “Tun-
P. K. Day, “Noise properties of superconducting coplanar waveg- able superconducting qubits with flux-independent coherence,”
uide microwave resonators,” Applied Physics Letters 90, 102507 Phys. Rev. Applied 8, 044003 (2017).
(2007). 70 J. M. Chow, J. M. Gambetta, A. W. Cross, S. T. Merkel,
58 J. Wenner, R. Barends, R. C. Bialczak, Y. Chen, J. Kelly, C. Rigetti, and M. Steffen, “Microwave-activated conditional-
E. Lucero, M. Mariantoni, A. Megrant, P. J. J. O’Malley, phase gate for superconducting qubits,” New Journal of Physics
D. Sank, A. Vainsencher, H. Wang, T. C. White, Y. Yin, 15, 115012 (2013).
J. Zhao, A. N. Cleland, and J. M. Martinis, “Surface loss simu- 71 R. Versluis, S. Poletto, N. Khammassi, B. Tarasinski, N. Haider,

lations of superconducting coplanar waveguide resonators,” Ap- D. J. Michalak, A. Bruno, K. Bertels, and L. DiCarlo, “Scal-
plied Physics Letters 99, 113513 (2011). able quantum circuit and control for a superconducting surface
59 L. J. Zeng, P. Krantz, S. Nik, P. Delsing, and E. Olsson, “The code,” Phys. Rev. Applied 8, 034021 (2017).
atomic details of the interfacial interaction between the bot- 72 T. P. Orlando, J. E. Mooij, L. Tian, C. H. van der Wal, L. S.

tom electrode of al/alox/al josephson junctions and hf-treated Levitov, S. Lloyd, and J. J. Mazo, “Superconducting persistent-
si substrates,” Journal of Applied Physics 117, 163915 (2015). current qubit,” Phys. Rev. B 60, 15398–15413 (1999).
60 L. J. Zeng, S. Nik, T. Greibe, P. Krantz, C. M. Wilson, P. Dels- 73 J. E. Mooij, T. P. Orlando, L. Levitov, L. Tian, C. H. van der

ing, and E. Olsson, “Direct observation of the thickness dis- Wal, and S. Lloyd, “Josephson persistent-current qubit,” Sci-
tribution of ultra thin alo x barriers in al/alo x /al josephson ence 285, 1036–1039 (1999).
junctions,” Journal of Physics D: Applied Physics 48, 395308 74 J. Q. You, J. S. Tsai, and F. Nori, “Hybridized solid-state qubit

(2015). in the charge-flux regime,” Phys. Rev. B 73, 014510 (2006).

61 G. Calusine, A. Melville, W. Woods, R. Das, C. Stull, 75 I. M. Pop, K. Geerlings, G. Catelani, R. J. Schoelkopf, L. I.

V. Bolkhovsky, D. Braje, D. Hover, D. K. Kim, X. Miloshi, Glazman, and M. H. Devoret, “Coherent suppression of elec-
D. Rosenberg, A. Sevi, J. L. Yoder, E. Dauler, and W. D. Oliver, tromagnetic dissipation due to superconducting quasiparticles,”
“Analysis and mitigation of interface losses in trenched super- Nature 508, 369 (2014).
conducting coplanar waveguide resonators,” Applied Physics 76 N. Earnest, S. Chakram, Y. Lu, N. Irons, R. K. Naik, N. Leung,

Letters 112, 062601 (2018). L. Ocola, D. A. Czaplewski, B. Baker, J. Lawrence, J. Koch,

62 F. Yan, S. Gustavsson, A. Kamal, J. Birenbaum, A. P. Sears, and D. I. Schuster, “Realization of a Λ system with metastable
D. Hover, T. J. Gudmundsen, D. Rosenberg, G. Samach, S. We- states of a capacitively shunted fluxonium,” Phys. Rev. Lett.
ber, J. L. Yoder, T. P. Orlando, J. Clarke, A. J. Kerman, and 120, 150504 (2018).
W. D. Oliver, “The flux qubit revisited to enhance coherence 77 Y.-H. Lin, L. B. Nguyen, N. Grabon, J. San Miguel, N. Pankra-

and reproducibility,” Nature Communications 7, 12964 (2016). tova, and V. E. Manucharyan, “Demonstration of protection of
63 D. C. McKay, S. Filipp, A. Mezzacapo, E. Magesan, J. M. Chow, a superconducting qubit from energy decay,” Phys. Rev. Lett.
and J. M. Gambetta, “Universal gate for fixed-frequency qubits 120, 150503 (2018).
via a tunable bus,” Phys. Rev. Applied 6, 064007 (2016). 78 J. Bylander, S. Gustavsson, F. Yan, F. Yoshihara, K. Harrabi,
64 L. DiCarlo, J. M. Chow, J. M. Gambetta, L. S. Bishop, B. R. G. Fitch, D. G. Cory, Y. Nakamura, J.-S. Tsai, and W. D.
Johnson, D. I. Schuster, J. Majer, A. Blais, L. Frunzio, S. M. Oliver, “Noise spectroscopy through dynamical decoupling with
Girvin, and R. J. Schoelkopf, “Demonstration of two-qubit al- a superconducting flux qubit,” Nature Physics 7, 565–570
gorithms with a superconducting quantum processor,” Nature (2011).
460, 240 (2009). 79 V. E. Manucharyan, J. Koch, L. I. Glazman, and M. H. Devoret,
65 R. Barends, J. Kelly, A. Megrant, A. Veitia, D. Sank, E. Jeffrey, “Fluxonium: Single cooper-pair circuit free of charge offsets,”
T. C. White, J. Mutus, A. G. Fowler, B. Campbell, Y. Chen, Science 326, 113–116 (2009).
Z. Chen, B. Chiaro, A. Dunsworth, C. Neill, P. O’Malley, 80 N. A. Masluk, I. M. Pop, A. Kamal, Z. K. Minev, and M. H.

P. Roushan, A. Vainsencher, J. Wenner, A. N. Korotkov, A. N. Devoret, “Microwave characterization of josephson junction ar-
Cleland, and J. M. Martinis, “Superconducting quantum cir- rays: Implementing a low loss superinductance,” Phys. Rev.
cuits at the surface code threshold for fault tolerance,” Nature Lett. 109, 137002 (2012).
508, 500 (2014). 81 V. E. Manucharyan, Superinductance, Ph.D. thesis (2012).
66 J. Kelly, R. Barends, a. G. Fowler, a. Megrant, E. Jeffrey, T. C. 82 J. R. Friedman and D. V. Averin, “Aharonov-casher-effect sup-

White, D. Sank, J. Y. Mutus, B. Campbell, Y. Chen, Z. Chen, pression of macroscopic tunneling of magnetic flux,” Phys. Rev.
B. Chiaro, a. Dunsworth, I.-C. Hoi, C. Neill, P. J. J. O’Malley, Lett. 88, 050403 (2002).
C. Quintana, P. Roushan, a. Vainsencher, J. Wenner, a. N. Cle- 83 A. J. Kerman, “Metastable superconducting qubit,” Phys. Rev.

land, and J. M. Martinis, “State preservation by repetitive error Lett. 104, 027002 (2010).
detection in a superconducting quantum circuit,” Nature 519, 84 P. Groszkowski, A. D. Paolo, A. L. Grimsmo, A. Blais, D. I.

66–69 (2015). Schuster, A. A. Houck, and J. Koch, “Coherence properties of

67 M. Reagor, C. B. Osborn, N. Tezak, A. Staley, G. Prawiroat- the 0-π qubit,” New Journal of Physics 20, 043053 (2018).
modjo, M. Scheer, N. Alidoust, E. A. Sete, N. Didier, M. P. 85 R. Barends, J. Kelly, A. Megrant, D. Sank, E. Jeffrey, Y. Chen,

da Silva, E. Acala, J. Angeles, A. Bestwick, M. Block, Y. Yin, B. Chiaro, J. Mutus, C. Neill, P. O’Malley, P. Roushan,
 58

J. Wenner, T. C. White, A. N. Cleland, and J. M. Martinis, K. Shain, W. Pfaff, Y. Chu, L. Frunzio, and R. J. Schoelkopf,
“Coherent josephson qubit suitable for scalable quantum inte- “An architecture for integrating planar and 3d cqed devices,”
grated circuits,” Phys. Rev. Lett. 111, 080502 (2013). Applied Physics Letters 109, 042601 (2016).
86 G. Burkard, R. H. Koch, and D. P. DiVincenzo, “Multilevel 103 M. Kounalakis, C. Dickel, A. Bruno, N. K. Langford, and G. A.

quantum description of decoherence in superconducting qubits,” Steele, “Tuneable hopping and nonlinear cross-Kerr interactions
Phys. Rev. B 69. in a high-coherence superconducting circuit,” npj Quantum In-
87 S. J. Weber, G. O. Samach, D. Hover, S. Gustavsson, D. K. Kim, formation 4, 38 (2018).
A. Melville, D. Rosenberg, A. P. Sears, F. Yan, J. L. Yoder, 104 Y. Chen, C. Neill, P. Roushan, N. Leung, M. Fang, R. Barends,

W. D. Oliver, and A. J. Kerman, “Coherent coupled qubits for J. Kelly, B. Campbell, Z. Chen, B. Chiaro, A. Dunsworth,
quantum annealing,” Phys. Rev. Applied 8, 014004 (2017). E. Jeffrey, A. Megrant, J. Y. Mutus, P. J. J. O’Malley, C. M.
88 M. W. Johnson, M. H. S. Amin, S. Gildert, T. Lanting, Quintana, D. Sank, A. Vainsencher, J. Wenner, T. C. White,
F. Hamze, N. Dickson, R. Harris, A. J. Berkley, J. Johansson, M. R. Geller, A. N. Cleland, and J. M. Martinis, “Qubit Ar-
P. Bunyk, E. M. Chapple, C. Enderud, J. P. Hilton, K. Karimi, chitecture with High Coherence and Fast Tunable Coupling,”
E. Ladizinsky, N. Ladizinsky, T. Oh, I. Perminov, C. Rich, M. C. Physical Review Letters 113, 220502 (2014).
Thom, E. Tolkacheva, C. J. S. Truncik, S. Uchaikin, J. Wang, 105 M. Roth, M. Ganzhorn, N. Moll, S. Filipp, G. Salis, and
B. Wilson, and G. Rose, “Quantum annealing with manufac- S. Schmidt, “Analysis of a parametrically driven exchange-type
tured spins,” Nature 473, 194–198 (2011). gate and a two-photon excitation gate between superconducting
89 J. Q. You, Y. Nakamura, and F. Nori, “Fast two-bit operations qubits,” Physical Review A 96, 062323 (2017).
in inductively coupled flux qubits,” Phys. Rev. B 71, 024532 106 N. Didier, E. A. Sete, M. P. da Silva, and C. Rigetti, “Ana-

(2005). lytical modeling of parametrically modulated transmon qubits,”

90 M. Grajcar, Y.-x. Liu, F. Nori, and A. M. Zagoskin, “Switch- Physical Review A 97, 022330 (2018).
able resonant coupling of flux qubits,” Phys. Rev. B 74, 172505 107 F. Yan, D. Campbell, P. Krantz, M. Kjaergaard, D. Kim, J. L.

(2006). Yoder, D. Hover, A. Sears, A. J. Kerman, T. P. Orlando, S. Gus-

91 A. O. Niskanen, K. Harrabi, F. Yoshihara, Y. Nakamura, tavsson, and W. D. Oliver, “Distinguishing coherent and ther-
S. Lloyd, and J. S. Tsai, “Quantum coherent tunable coupling of mal photon noise in a circuit quantum electrodynamical sys-
superconducting qubits.” Science (New York, N.Y.) 316, 723–6 tem,” Phys. Rev. Lett. 120, 260504 (2018).
(2007). 108 N. Didier, J. Bourassa, and A. Blais, “Fast quantum nondemo-
92 S. Ashhab, A. O. Niskanen, K. Harrabi, Y. Nakamura, T. Pi- lition readout by parametric modulation of longitudinal qubit-
cot, P. C. de Groot, C. J. P. M. Harmans, J. E. Mooij, and oscillator interaction,” Phys. Rev. Lett. 115, 203601 (2015).
F. Nori, “Interqubit coupling mediated by a high-excitation- 109 H. Ball, W. D. Oliver, and M. J. Biercuk, “The role of master

energy quantum object,” Phys. Rev. B 77, 014510 (2008). clock stability in quantum information processing,” Npj Quan-
93 F. Yoshihara, T. Fuse, S. Ashhab, K. Kakuyanagi, S. Saito, and tum Information 2, 16033 (2016).
K. Semba, “Superconducting qubit-oscillator circuit beyond the 110 R. K. Wangsness, “Sublattice effects in magnetic resonance,”

ultrastrong-coupling regime,” Nature Physics 13, 44 (2016). Phys. Rev. 91, 1085–1091 (1953).
94 T. Niemczyk, F. Deppe, H. Huebl, E. P. Menzel, F. Hocke, M. J. 111 F. Bloch, “Generalized theory of relaxation,” Phys. Rev. 105,

Schwarz, J. J. Garcia-Ripoll, D. Zueco, T. Hümmer, E. Solano, 1206–1222 (1957).

A. Marx, and R. Gross, “Circuit quantum electrodynamics in 112 R. A. G., “On the theory of relaxation processes,” IBM Journal

the ultrastrong-coupling regime,” Nature Physics 6, 772 (2010). of Research and Development 1, 19–31 (1957).
95 Y. Lu, S. Chakram, N. Leung, N. Earnest, R. K. Naik, Z. Huang, 113 G. Ithier, Manipulation, readout and analysis of the decoherence

P. Groszkowski, E. Kapit, J. Koch, and D. I. Schuster, “Univer- ofa superconducting quantum bit, Ph.D. thesis (2005).
sal stabilization of a parametrically coupled qubit,” Phys. Rev. 114 G. Ithier, E. Collin, P. Joyez, P. J. Meeson, D. Vion, D. Es-

Lett. 119, 150502 (2017). teve, F. Chiarello, A. Shnirman, Y. Makhlin, J. Schriefl, and
96 A. Wallraff, D. I. Schuster, A. Blais, L. Frunzio, J. Majer, S. Ku- G. Schön, “Decoherence in a superconducting quantum bit cir-
mar, S. M. Girvin, and R. J. Schoelkopf, “Strong coupling of a cuit,” Phys. Rev. B 72, 134519 (2005).
single photon to a superconducting qubit using circuit quantum 115 N. F. Ramsey, “A molecular beam resonance method with sep-

electrodynamics,” Nature 431, 162–166 (2004). arated oscillating fields,” Phys. Rev. 78, 695–699 (1950).
97 M. Sandberg, C. M. Wilson, F. Persson, T. Bauch, G. Johans- 116 E. L. Hahn, “Spin echoes,” Phys. Rev. 80, 580–594 (1950).

son, V. Shumeiko, T. Duty, and P. Delsing, “Tuning the field in 117 M. Kjaergaard, P. Krantz, T. Larsen, M. Kimchi-Schwarz,

a microwave resonator faster than the photon lifetime,” Applied D. Rosenberg, J. Yoder, D. Kim, S. Gustavsson, and W. D.
Physics Letters 92 (2008), 10.1063/1.2929367, arXiv:0801.2479. Oliver, “Unpublished,” Unpublished.
98 M. Pierre, I.-M. Svensson, S. Raman Sathyamoorthy, G. Jo- 118 A. Siegman, Lasers (University Science Books, 1986).

hansson, and P. Delsing, “Storage and on-demand release of 119 D. M. Berns, W. D. Oliver, S. O. Valenzuela, A. V. Shytov,

microwaves using superconducting resonators with tunable cou- K. K. Berggren, L. S. Levitov, and T. P. Orlando, “Coher-
pling,” Applied Physics Letters 104, 232604 (2014). ent quasiclassical dynamics of a persistent current qubit,” Phys.
99 Y. Yin, Y. Chen, D. Sank, P. J. J. O’Malley, T. C. White, Rev. Lett. 97, 150502 (2006).
R. Barends, J. Kelly, E. Lucero, M. Mariantoni, A. Megrant, 120 G. Catelani, J. Koch, L. Frunzio, R. J. Schoelkopf, M. H. De-

C. Neill, A. Vainsencher, J. Wenner, A. N. Korotkov, A. N. voret, and L. I. Glazman, “Quasiparticle relaxation of super-
Cleland, and J. M. Martinis, “Catch and release of microwave conducting qubits in the presence of flux,” Phys. Rev. Lett. 106,
photon states,” Phys. Rev. Lett. 110, 107001 (2013). 077002 (2011).
100 N. Ofek, A. Petrenko, R. Heeres, P. Reinhold, Z. Leghtas, 121 G. Catelani, S. E. Nigg, S. M. Girvin, R. J. Schoelkopf, and

B. Vlastakis, Y. Liu, L. Frunzio, S. M. Girvin, L. Jiang, M. Mir- L. I. Glazman, “Decoherence of superconducting qubits caused
rahimi, M. H. Devoret, and R. J. Schoelkopf, “Extending the by quasiparticle tunneling,” Phys. Rev. B 86, 184514 (2012).
lifetime of a quantum bit with error correction in superconduct- 122 S. Gustavsson, F. Yan, G. Catelani, J. Bylander, A. Kamal,

ing circuits,” Nature 536, 441–445 (2016). J. Birenbaum, D. Hover, D. Rosenberg, G. Samach, A. P. Sears,
101 C. Wang, Y. Y. Gao, P. Reinhold, R. W. Heeres, N. Ofek, S. J. Weber, J. L. Yoder, J. Clarke, A. J. Kerman, F. Yoshihara,
K. Chou, C. Axline, M. Reagor, J. Blumoff, K. M. Sliwa, Y. Nakamura, T. P. Orlando, and W. D. Oliver, “Suppressing
L. Frunzio, S. M. Girvin, L. Jiang, M. Mirrahimi, M. H. De- relaxation in superconducting qubits by quasiparticle pumping,”
voret, and R. J. Schoelkopf, “A schrödinger cat living in two Science (2016), 10.1126/science.aah5844.
boxes,” Science 352, 1087–1091 (2016). 123 G. Falci, A. D’Arrigo, A. Mastellone, and E. Paladino, “Initial
102 C. Axline, M. Reagor, R. Heeres, P. Reinhold, C. Wang, decoherence in solid state qubits,” Phys. Rev. Lett. 94, 167002
 59

(2005). stration of fidelity improvement using dynamical decoupling

124 E. Paladino, Y. M. Galperin, G. Falci, and B. L. Altshuler, with superconducting qubits,” Phys. Rev. Lett. 121, 220502
“1/f noise: Implications for solid-state quantum information,” (2018).
Rev. Mod. Phys. 86, 361–418 (2014). 144 C. M. Quintana, Y. Chen, D. Sank, A. G. Petukhov, T. C.
125 H. Nyquist, “Thermal agitation of electric charge in conduc- White, D. Kafri, B. Chiaro, A. Megrant, R. Barends, B. Camp-
tors,” Phys. Rev. 32, 110–113 (1928). bell, Z. Chen, A. Dunsworth, A. G. Fowler, R. Graff, E. Jef-
126 J. B. Johnson, “Thermal agitation of electricity in conductors,” frey, J. Kelly, E. Lucero, J. Y. Mutus, M. Neeley, C. Neill,
Phys. Rev. 32, 97–109 (1928). P. J. J. O’Malley, P. Roushan, A. Shabani, V. N. Smelyanskiy,
127 H. A. Haus, Electromagnetic Noise and Quantum Optical Mea- A. Vainsencher, J. Wenner, H. Neven, and J. M. Martinis, “Ob-
surements, 1st ed. (pringer-Verlag Berlin Heidelberg, Berlin, servation of classical-quantum crossover of 1/f flux noise and its
Germany, 2000). paramagnetic temperature dependence,” Phys. Rev. Lett. 118,
128 A. A. Clerk, M. H. Devoret, S. M. Girvin, F. Marquardt, and 057702 (2017).
R. J. Schoelkopf, “Introduction to quantum noise, measurement, 145 P. Kumar, S. Sendelbach, M. A. Beck, J. W. Freeland, Z. Wang,

and amplification,” Rev. Mod. Phys. 82, 1155–1208 (2010). H. Wang, C. C. Yu, R. Q. Wu, D. P. Pappas, and R. Mc-
129 S. Gustavsson, J. Bylander, F. Yan, W. D. Oliver, F. Yoshihara, Dermott, “Origin and reduction of 1/f magnetic flux noise in
and Y. Nakamura, “Noise correlations in a flux qubit with tun- superconducting devices,” Phys. Rev. Applied 6, 041001 (2016).
able tunnel coupling,” Phys. Rev. B 84, 014525 (2011). 146 S. E. de Graaf, A. A. Adamyan, T. Lindström, D. Erts, S. E.
130 F. Yoshihara, Y. Nakamura, F. Yan, S. Gustavsson, J. Bylander, Kubatkin, A. Y. Tzalenchuk, and A. V. Danilov, “Direct iden-
W. D. Oliver, and J.-S. Tsai, “Flux qubit noise spectroscopy tification of dilute surface spins on al2 o3 : Origin of flux noise in
using rabi oscillations under strong driving conditions,” Phys. quantum circuits,” Phys. Rev. Lett. 118, 057703 (2017).
Rev. B 89, 020503 (2014). 147 D. I. Schuster, A. Wallraff, A. Blais, L. Frunzio, R.-S. Huang,
131 F. Yan, J. Bylander, S. Gustavsson, F. Yoshihara, K. Harrabi, J. Majer, S. M. Girvin, and R. J. Schoelkopf, “ac stark shift
D. G. Cory, T. P. Orlando, Y. Nakamura, J.-S. Tsai, and W. D. and dephasing of a superconducting qubit strongly coupled to
Oliver, “Spectroscopy of low-frequency noise and its tempera- a cavity field,” Phys. Rev. Lett. 94, 123602 (2005).
ture dependence in a superconducting qubit,” Phys. Rev. B 85, 148 J.-H. Yeh, J. LeFebvre, S. Premaratne, F. C. Wellstood, and

174521 (2012). B. S. Palmer, “Microwave attenuators for use with quantum

132 F. Yan, S. Gustavsson, J. Bylander, X. Jin, F. Yoshihara, devices below 100 mk,” Journal of Applied Physics 121, 224501
D. G. Cory, Y. Nakamura, T. P. Orlando, and W. D. Oliver, (2017).
“Rotating-frame relaxation as a noise spectrum analyser of a su- 149 X. Y. Jin, A. Kamal, A. P. Sears, T. Gudmundsen, D. Hover,

perconducting qubit undergoing driven evolution,” Nature Com- J. Miloshi, R. Slattery, F. Yan, J. Yoder, T. P. Orlando, S. Gus-
munications 4, 2337 (2013). tavsson, and W. D. Oliver, “Thermal and residual excited-
133 Y. Sung, F. Beaudoin, L. Norris, F. Yan, D. Kim, J. Qiu, U. von state population in a 3d transmon qubit,” Phys. Rev. Lett. 114,
Lüepke, J. L. Yoder, T. Orlando, L. Viola, S. Gustavsson, and 240501 (2015).
W. Oliver, “Non-Gaussian noise spectroscopy with a supercon- 150 K. Serniak, M. Hays, G. de Lange, S. Diamond, S. Shankar,

ducting qubit sensor,” arXiv:1903.01043 (2019). L. D. Burkhart, L. Frunzio, M. Houzet, and M. H. Devoret,
134 C. Wang, C. Axline, Y. Y. Gao, T. Brecht, Y. Chu, L. Frunzio, “Hot nonequilibrium quasiparticles in transmon qubits,” Phys.
M. H. Devoret, and R. J. Schoelkopf, “Surface participation Rev. Lett. 121, 157701 (2018).
and dielectric loss in superconducting qubits,” Applied Physics 151 J. Q. You, X. Hu, S. Ashhab, and F. Nori, “Low-decoherence

Letters 107, 162601 (2015). flux qubit,” Phys. Rev. B 75, 140515 (2007).
135 O. Dial, D. T. McClure, S. Poletto, G. A. Keefe, M. B. Rothwell, 152 A. J. Kerman and W. D. Oliver, “High-fidelity quantum opera-

J. M. Gambetta, D. W. Abraham, J. M. Chow, and M. Steffen, tions on superconducting qubits in the presence of noise,” Phys.
“Bulk and surface loss in superconducting transmon qubits,” Rev. Lett. 101, 070501 (2008).
Superconductor Science and Technology 29, 044001 (2016). 153 W. D. Oliver and P. B. Welander, “Materials in superconducting
136 O. Astafiev, Y. A. Pashkin, Y. Nakamura, T. Yamamoto, and quantum bits,” MRS Bulletin 38, 816–825 (2013).
J. S. Tsai, “Quantum noise in the josephson charge qubit,” Phys. 154 J. M. Martinis, S. Nam, J. Aumentado, K. M. Lang, and
Rev. Lett. 93, 267007 (2004). C. Urbina, “Decoherence of a superconducting qubit due to bias
137 R. H. Koch, D. P. DiVincenzo, and J. Clarke, “Model for 1/f noise,” Phys. Rev. B 67, 094510 (2003).
flux noise in squids and qubits,” Phys. Rev. Lett. 98, 267003 155 G. S. Uhrig, “Keeping a quantum bit alive by optimized π-pulse

(2007). sequences,” Phys. Rev. Lett. 98, 100504 (2007).

138 F. C. Wellstood, C. Urbina, and J. Clarke, “Low-frequency 156 L. Cywinski, R. M. Lutchyn, C. P. Nave, and S. Das Sarma,

noise in dc superconducting quantum interference devices below “How to enhance dephasing time in superconducting qubits,”
1 k,” Applied Physics Letters 50, 772–774 (1987). Phys. Rev. B 77, 174509 (2008).
139 D. H. Slichter, R. Vijay, S. J. Weber, S. Boutin, M. Bois- 157 M. J. Biercuk, H. Uys, A. P. VanDevender, N. Shiga, W. M.

sonneault, J. M. Gambetta, A. Blais, and I. Siddiqi, Itano, and J. J. Bollinger, “Experimental uhrig dynamical de-
“Measurement-induced qubit state mixing in circuit qed from coupling using trapped ions,” Phys. Rev. A 79, 062324 (2009).
up-converted dephasing noise,” Phys. Rev. Lett. 109, 153601 158 H. Uys, M. J. Biercuk, and J. J. Bollinger, “Optimized noise

(2012). filtration through dynamical decoupling,” Phys. Rev. Lett. 103,

140 G. Falci, A. D’Arrigo, A. Mastellone, and E. Paladino, “Dy- 040501 (2009).
namical suppression of telegraph and 1/f noise due to quantum 159 H. Y. Carr and E. M. Purcell, “Effects of diffusion on free pre-

bistable fluctuators,” Phys. Rev. A 70, 040101 (2004). cession in nuclear magnetic resonance experiments,” Phys. Rev.
141 R. C. Bialczak, R. McDermott, M. Ansmann, M. Hofheinz, 94, 630–638 (1954).
N. Katz, E. Lucero, M. Neeley, A. D. O’Connell, H. Wang, 160 S. Meiboom and D. Gill, “Modified spin-echo method for mea-

A. N. Cleland, and J. M. Martinis, “1/f flux noise in josephson suring nuclear relaxation times,” Review of Scientific Instru-
phase qubits,” Phys. Rev. Lett. 99, 187006 (2007). ments 29, 688 (1954).
142 S. M. Anton, C. Müller, J. S. Birenbaum, S. R. O’Kelley, A. D. 161 S. Gustavsson, J. Bylander, F. Yan, P. Forn-Dı́az,
Fefferman, D. S. Golubev, G. C. Hilton, H.-M. Cho, K. D. Irwin, V. Bolkhovsky, D. Braje, G. Fitch, K. Harrabi, D. Lennon,
F. C. Wellstood, G. Schön, A. Shnirman, and J. Clarke, “Pure J. Miloshi, P. Murphy, R. Slattery, S. Spector, B. Turek,
dephasing in flux qubits due to flux noise with spectral density T. Weir, P. B. Welander, F. Yoshihara, D. G. Cory, Y. Naka-
scaling as 1/f α ,” Phys. Rev. B 85, 224505 (2012). mura, T. P. Orlando, and W. D. Oliver, “Driven dynamics
143 B. Pokharel, N. Anand, B. Fortman, and D. A. Lidar, “Demon- and rotary echo of a qubit tunably coupled to a harmonic
 60

oscillator,” Phys. Rev. Lett. 108, 170503 (2012). “Elementary gates for quantum computation,” Physical Review
162 S. Gustavsson, F. Yan, J. Bylander, F. Yoshihara, Y. Nakamura, A 52, 3457–3467 (1995).
T. P. Orlando, and W. D. Oliver, “Dynamical decoupling and 180 C. P. Williams, Explorations in Quantum Computing, 2nd ed.

dephasing in interacting two-level systems,” Phys. Rev. Lett. (Springer Publishing Company, Incorporated, 2008).
109, 010502 (2012). 181 Y. A. Kitaev, “Quantum computations: algorithms and error
163 D. T. Sank, Fast, accurate state measurement in superconduct- correction,” Russian Mathematical Surveys 52, 1191 (1997).
ing qubits, Ph.D. thesis (2014). 182 C. M. Dawson and M. A. Nielsen, “The Solovay-Kitaev Algo-
164 M. Loretz, J. M. Boss, T. Rosskopf, H. J. Mamin, D. Rugar, rithm,” Quantum Info. Comput. 6, 81–95 (2006).
and C. L. Degen, “Spurious harmonic response of multipulse 183 F. T. Chong, D. Franklin, and M. Martonosi, “Programming

quantum sensing sequences,” Phys. Rev. X 5, 021009 (2015). languages and compiler design for realistic quantum hardware,”
165 D. Slepian and H. O. Pollak, “Prolate spheroidal wave functions, Nature 549, 180 (2017).
fourier analysis and uncertainty — i,” Bell System Technical 184 E. T. Campbell and M. Howard, “Unified framework for magic

Journal 40, 43–63 (1961). state distillation and multiqubit gate synthesis with reduced
166 R. Schoelkopf, A. Clerk, S. Girvin, K. Lehnert, and M. Devoret, resource cost,” Physical Review A 95, 022316 (2017).
Quantum Noise in Mesoscopic Physics, Qubits as Spectrome- 185 A. Paler, I. Polian, K. Nemoto, and S. J. Devitt, “Fault-
ters of Quantum Noise (Part of the NATO Science Series book tolerant, high-level quantum circuits: form, compilation and
series (NAII, volume 97), 2002) pp. 175 – 203. description,” Quantum Science and Technology 2, 25003 (2017).
167 J. M. Sage, V. Bolkhovsky, W. D. Oliver, B. Turek, and P. B. 186 U. Vool and M. H. Devoret, “Introduction to quantum electro-

Welander, “Study of loss in superconducting coplanar waveguide magnetic circuits,” International Journal of Circuit Theory and
resonators,” Journal of Applied Physics 109, 063915 (2011). Applications 45, 897–934 (2017).
168 C. M. Quintana, A. Megrant, Z. Chen, A. Dunsworth, 187 S. M. Girvin, “Circuit QED: superconducting qubits coupled

B. Chiaro, R. Barends, B. Campbell, Y. Chen, I.-C. Hoi, to microwave photons,” in Quantum Machines: Measurement
E. Jeffrey, J. Kelly, J. Y. Mutus, P. J. J. O’Malley, C. Neill, and Control of Engineered Quantum Systems (Oxford Univer-
P. Roushan, D. Sank, A. Vainsencher, J. Wenner, T. C. White, sity Press, 2014) pp. 113–256.
A. N. Cleland, and J. M. Martinis, “Characterization and re- 188 N. K. Langford, “Circuit QED - Lecture Notes,” ArXiv e-prints

duction of microfabrication-induced decoherence in supercon- (2013), arXiv:1310.1897 [quant-ph].

ducting quantum circuits,” Applied Physics Letters 105, 062601 189 V. Vesterinen, O. P. Saira, A. Bruno, and L. DiCarlo, “Mit-

(2014). igating information leakage in a crowded spectrum of weakly

169 A. Kamal, J. L. Yoder, F. Yan, T. J. Gudmundsen, anharmonic qubits,” (2014), arXiv:1405.0450.
D. Hover, A. P. Sears, P. Welander, T. P. Orlando, S. Gus- 190 S. Asaad, C. Dickel, N. K. Langford, S. Poletto, A. Bruno, M. A.

tavsson, and W. D. Oliver, “Improved superconducting Rol, D. Deurloo, and L. DiCarlo, “Independent, extensible
qubit coherence with high-temperature substrate annealing,” control of same-frequency superconducting qubits by selective
arXiv:arXiv:1606.09262. broadcasting,” npj Quantum Information 2, 16029 (2016).
170 S. M. Anton, J. S. Birenbaum, S. R. O’Kelley, V. Bolkhovsky, 191 C. A. Ryan, B. R. Johnson, D. Ristè, B. Donovan, and T. A.

D. A. Braje, G. Fitch, M. Neeley, G. C. Hilton, H.-M. Cho, Ohki, “Hardware for dynamic quantum computing,” Review of
K. D. Irwin, F. C. Wellstood, W. D. Oliver, A. Shnirman, and Scientific Instruments 88, 104703 (2017).
J. Clarke, “Magnetic flux noise in dc squids: Temperature and 192 D. C. McKay, C. J. Wood, S. Sheldon, J. M. Chow, and J. M.

geometry dependence,” Phys. Rev. Lett. 110, 147002 (2013). Gambetta, “Efficient $Z$ gates for quantum computing,” Phys.
171 S. Sendelbach, D. Hover, A. Kittel, M. Mück, J. M. Martinis, Rev. A 96, 22330 (2017).
and R. McDermott, “Magnetism in squids at millikelvin tem- 193 J. Johansson, P. Nation, and F. Nori, “QuTiP 2: A Python

peratures,” Phys. Rev. Lett. 100, 227006 (2008). framework for the dynamics of open quantum systems,” Com-
172 R.-P. Riwar, A. Hosseinkhani, L. D. Burkhart, Y. Y. Gao, R. J. puter Physics Communications 184, 1234–1240 (2013).
Schoelkopf, L. I. Glazman, and G. Catelani, “Normal-metal 194 Z. Chen, Metrology of Quantum Control and Measurement in

quasiparticle traps for superconducting qubits,” Phys. Rev. B Superconducting Qubits, Ph.D. thesis (2018).
94, 104516 (2016). 195 F. Motzoi, J. M. Gambetta, P. Rebentrost, and F. K. Wilhelm,
173 S. Krinner, S. Storz, P. Kurpiers, P. Magnard, J. Heinsoo, “Simple Pulses for Elimination of Leakage in Weakly Nonlinear
R. Keller, J. Luetolf, C. Eichler, and A. Wallraff, “Engineer- Qubits,” Physical Review Letters 103, 110501 (2009).
ing cryogenic setups for 100-qubit scale superconducting circuit 196 J. M. Gambetta, F. Motzoi, S. T. Merkel, and F. K. Wilhelm,

systems,” arXiv:arXiv:1806.07862. “Analytic control methods for high-fidelity unitary operations

174 R. Barends, J. Wenner, M. Lenander, Y. Chen, R. C. Bial- in a weakly nonlinear oscillator,” Physical Review A 83, 012308
czak, J. Kelly, E. Lucero, P. O’Malley, M. Mariantoni, D. Sank, (2011).
H. Wang, T. C. White, Y. Yin, J. Zhao, A. N. Cleland, J. M. 197 A. De, “Fast Quantum Control for Weakly Nonlinear

Martinis, and J. J. A. Baselmans, “Minimizing quasiparticle Qubits: On Two-Quadrature Adiabatic Gates,” (2015),
generation from stray infrared light in superconducting quan- arXiv:1509.07905.
tum circuits,” Applied Physics Letters 99, 113507 (2011). 198 F. Motzoi and F. K. Wilhelm, “Improving frequency selection
175 A. D. Córcoles, J. M. Chow, J. M. Gambetta, C. Rigetti, J. R. of driven pulses using derivative-based transition suppression,”
Rozen, G. A. Keefe, M. Beth Rothwell, M. B. Ketchen, and Physical Review A 88, 062318 (2013).
M. Steffen, “Protecting superconducting qubits from radiation,” 199 Z. Chen, J. Kelly, C. Quintana, R. Barends, B. Campbell,

Applied Physics Letters 99, 181906 (2011). Y. Chen, B. Chiaro, A. Dunsworth, A. Fowler, E. Lucero, E. Jef-
176 Z. Wang, S. Shankar, Z. Minev, P. Campagne-Ibarcq, A. Narla, frey, A. Megrant, J. Mutus, M. Neeley, C. Neill, P. O’Malley,
and M. Devoret, “Cavity attenuators for superconducting P. Roushan, D. Sank, A. Vainsencher, J. Wenner, T. White,
qubits,” Phys. Rev. Applied 11, 014031 (2019). A. Korotkov, and J. M. Martinis, “Measuring and Suppressing
177 W. D. Oliver, I. L. Chuang, A. W. Harrow, and P. Shor, Quantum State Leakage in a Superconducting Qubit,” Physical
“Mit xpro quantum curriculum,” quantumcurriculum.mit.edu Review Letters 116, 020501 (2016).
(2018). 200 J. M. Chow, L. DiCarlo, J. M. Gambetta, F. Motzoi, L. Frunzio,
178 M. A. Nielsen and I. L. Chuang, Quantum Computation and S. M. Girvin, and R. J. Schoelkopf, “Optimized driving of su-
Quantum Information: 10th Anniversary Edition, 10th ed. perconducting artificial atoms for improved single-qubit gates,”
(Cambridge University Press, New York, NY, USA, 2011). Physical Review A 82, 040305 (2010).
179 A. Barenco, C. H. Bennett, R. Cleve, D. P. DiVincenzo, N. Mar- 201 E. Lucero, J. Kelly, R. C. Bialczak, M. Lenander,
golus, P. Shor, T. Sleator, J. A. Smolin, and H. Weinfurter, M. Mariantoni, M. Neeley, A. D. O’Connell, D. Sank, H. Wang,
 61

M. Weides, J. Wenner, T. Yamamoto, A. N. Cleland, and J. M. cal Review A 69, 062320 (2004).
Martinis, “Reduced phase error through optimized control of a 216 J. Majer, J. M. Chow, J. M. Gambetta, J. Koch, B. R. John-
superconducting qubit,” Physical Review A 82, 042339 (2010). son, J. A. Schreier, L. Frunzio, D. I. Schuster, A. A. Houck,
202 S. Gustavsson, O. Zwier, J. Bylander, F. Yan, F. Yoshihara, A. Wallraff, A. Blais, M. H. Devoret, S. M. Girvin, and R. J.
Y. Nakamura, T. P. Orlando, and W. D. Oliver, “Improv- Schoelkopf, “Coupling superconducting qubits via a cavity bus,”
ing Quantum Gate Fidelities by Using a Qubit to Measure Nature 449, 443–447 (2007).
Microwave Pulse Distortions,” Physical Review Letters 110, 217 G. Wendin and V. S. Shumeiko, “Quantum bits with Josephson

040502 (2013). junctions (review article),” Low Temperature Physics 33, 724–
203 S. Sheldon, E. Magesan, J. M. Chow, and J. M. Gambetta, 744 (2007).
“Procedure for systematically tuning up cross-talk in the cross- 218 N. Schuch and J. Siewert, “Natural two-qubit gate for quantum

resonance gate,” Physical Review A 93, 060302 (2016). computation using the XY interaction,” Physical Review A 67,
204 S. Sheldon, L. S. Bishop, E. Magesan, S. Filipp, J. M. Chow, 032301 (2003).
and J. M. Gambetta, “Characterizing errors on qubit operations 219 M. Steffen, M. Ansmann, R. C. Bialczak, N. Katz, E. Lucero,

via iterative randomized benchmarking,” Physical Review A 93, R. McDermott, M. Neeley, E. M. Weig, A. N. Cleland, and J. M.
012301 (2016). Martinis, “Measurement of the entanglement of two supercon-
205 M. Rol, C. Bultink, T. O’Brien, S. de Jong, L. Theis, X. Fu, ducting qubits via state tomography.” Science (New York, N.Y.)
F. Luthi, R. Vermeulen, J. de Sterke, A. Bruno, D. Deurloo, 313, 1423–5 (2006).
R. Schouten, F. Wilhelm, and L. DiCarlo, “Restless Tuneup of 220 R. C. Bialczak, M. Ansmann, M. Hofheinz, E. Lucero, M. Nee-

High-Fidelity Qubit Gates,” Physical Review Applied 7, 041001 ley, A. D. O’Connell, D. Sank, H. Wang, J. Wenner, M. Steffen,
(2017). A. N. Cleland, and J. M. Martinis, “Quantum process tomogra-
206 R. Schutjens, F. A. Dagga, D. J. Egger, and F. K. Wilhelm, phy of a universal entangling gate implemented with Josephson
“Single-qubit gates in frequency-crowded transmon systems,” phase qubits,” Nature Physics 6, 409–413 (2010).
Physical Review A 88, 052330 (2013). 221 M. Neeley, R. C. Bialczak, M. Lenander, E. Lucero,
207 L. S. Theis, F. Motzoi, and F. K. Wilhelm, “Simultaneous gates M. Mariantoni, A. D. O’Connell, D. Sank, H. Wang, M. Weides,
in frequency-crowded multilevel systems using fast, robust, an- J. Wenner, Y. Yin, T. Yamamoto, A. N. Cleland, and J. M.
alytic control shapes,” Physical Review A 93, 012324 (2016). Martinis, “Generation of three-qubit entangled states using su-
208 S. H. W. van der Ploeg, A. Izmalkov, A. M. van den Brink, perconducting phase qubits,” Nature 467, 570–573 (2010).
U. Hübner, M. Grajcar, E. Il’ichev, H.-G. Meyer, and A. M. 222 A. Dewes, F. R. Ong, V. Schmitt, R. Lauro, N. Boulant,

Zagoskin, “Controllable Coupling of Superconducting Flux P. Bertet, D. Vion, and D. Esteve, “Characterization of a Two-
Qubits,” Physical Review Letters 98, 057004 (2007). Transmon Processor with Individual Single-Shot Qubit Read-
209 M. S. Allman, F. Altomare, J. D. Whittaker, K. Cicak, D. Li, out,” Physical Review Letters 108, 057002 (2012).
A. Sirois, J. Strong, J. D. Teufel, and R. W. Simmonds, “rf- 223 Y. Salathé, M. Mondal, M. Oppliger, J. Heinsoo, P. Kurpiers,

SQUID-Mediated Coherent Tunable Coupling between a Super- A. Potočnik, A. Mezzacapo, U. Las Heras, L. Lamata, E. Solano,
conducting Phase Qubit and a Lumped-Element Resonator,” S. Filipp, and A. Wallraff, “Digital Quantum Simulation of
Physical Review Letters 104, 177004 (2010). Spin Models with Circuit Quantum Electrodynamics,” Physical
210 S. J. Srinivasan, A. J. Hoffman, J. M. Gambetta, and A. A. Review X 5, 021027 (2015).
Houck, “Tunable Coupling in Circuit Quantum Electrodynam- 224 L. Casparis, M. R. Connolly, M. Kjaergaard, N. J. Pearson,

ics Using a Superconducting Charge Qubit with a V -Shaped A. Kringhøj, T. W. Larsen, F. Kuemmeth, T. Wang, C. Thomas,
Energy Level Diagram,” Physical Review Letters 106, 083601 S. Gronin, G. C. Gardner, M. J. Manfra, C. M. Marcus, and
(2011). K. D. Petersson, “Superconducting gatemon qubit based on a
211 A. J. Sirois, M. A. Castellanos-Beltran, M. P. DeFeo, L. Ran- proximitized two-dimensional electron gas,” Nature Nanotech-
zani, F. Lecocq, R. W. Simmonds, J. D. Teufel, and J. Au- nology , 1 (2018).
mentado, “Coherent-state storage and retrieval between super- 225 U. L. Heras, A. Mezzacapo, L. Lamata, S. Filipp, A. Wallraff,

conducting cavities using parametric frequency conversion,” Ap- and E. Solano, “Digital Quantum Simulation of Spin Systems
plied Physics Letters 106, 172603 (2015). in Superconducting Circuits,” Physical Review Letters 112,
212 S. Caldwell, N. Didier, C. A. Ryan, E. A. Sete, A. Hudson, 200501 (2014).
P. Karalekas, R. Manenti, M. Reagor, M. P. da Silva, R. Sin- 226 C. Song, K. Xu, W. Liu, C.-p. Yang, S.-B. Zheng, H. Deng,

clair, E. Acala, N. Alidoust, J. Angeles, A. Bestwick, M. Block, Q. Xie, K. Huang, Q. Guo, L. Zhang, P. Zhang, D. Xu, D. Zheng,
B. Bloom, A. Bradley, C. Bui, L. Capelluto, R. Chilcott, J. Cor- X. Zhu, H. Wang, Y.-A. Chen, C.-Y. Lu, S. Han, and J.-W. Pan,
dova, G. Crossman, M. Curtis, S. Deshpande, T. E. Bouayadi, “10-Qubit Entanglement and Parallel Logic Operations with a
D. Girshovich, S. Hong, K. Kuang, M. Lenihan, T. Manning, Superconducting Circuit,” Physical Review Letters 119, 180511
A. Marchenkov, J. Marshall, R. Maydra, Y. Mohan, W. O’Brien, (2017).
C. Osborn, J. Otterbach, A. Papageorge, J. P. Paquette, M. Pel- 227 F. W. Strauch, P. R. Johnson, A. J. Dragt, C. J. Lobb, J. R.

string, A. Polloreno, G. Prawiroatmodjo, V. Rawat, R. Renzas, Anderson, and F. C. Wellstood, “Quantum Logic Gates for
N. Rubin, D. Russell, M. Rust, D. Scarabelli, M. Scheer, M. Sel- Coupled Superconducting Phase Qubits,” Physical Review Let-
vanayagam, R. Smith, A. Staley, M. Suska, N. Tezak, D. C. ters 91, 167005 (2003).
Thompson, T. W. To, M. Vahidpour, N. Vodrahalli, T. Why- 228 J. Kelly, R. Barends, B. Campbell, Y. Chen, Z. Chen, B. Chiaro,

land, K. Yadav, W. Zeng, and C. Rigetti, “Parametrically A. Dunsworth, A. Fowler, I.-C. Hoi, E. Jeffrey, A. Megrant,
Activated Entangling Gates Using Transmon Qubits,” (2017), J. Mutus, C. Neill, P. O’Malley, C. Quintana, P. Roushan,
arXiv:1706.06562. D. Sank, A. Vainsencher, J. Wenner, T. White, A. Cleland, and
213 L. Casparis, N. J. Pearson, A. Kringhøj, T. W. Larsen, F. Kuem- J. M. Martinis, “Optimal Quantum Control Using Randomized
meth, J. Nygård, P. Krogstrup, K. D. Petersson, and C. M. Benchmarking,” Physical Review Letters 112, 240504 (2014).
Marcus, “Voltage-Controlled Superconducting Quantum Bus,” 229 W. D. Oliver, Y. Yu, J. C. Lee, K. K. Berggren, L. S. Levitov,

(2018), arXiv:1802.01327. and T. P. Orlando, “Mach-zehnder interferometry in a strongly

214 N. Didier, E. A. Sete, M. P. da Silva, and C. Rigetti, “Ana- driven superconducting qubit,” Science 310, 1653–1657 (2005).
lytical modeling of parametrically modulated transmon qubits,” 230 S. O. Valenzuela, W. D. Oliver, D. M. Berns, K. K. Berggren,

Physical Review A 97, 022330 (2018). L. S. Levitov, and T. P. Orlando, “Microwave-induced cooling
215 A. Blais, R.-S. Huang, A. Wallraff, S. Girvin, and R. Schoelkopf, of a superconducting qubit,” Science 314, 1589–1592 (2006).
“Cavity quantum electrodynamics for superconducting electri- 231 S. Ashhab, J. R. Johansson, A. M. Zagoskin, and F. Nori, “Two-

cal circuits: An architecture for quantum computation,” Physi- level systems driven by large-amplitude fields,” Phys. Rev. A 75,
 62

063414 (2007). ner, and J. M. Martinis, “Scalable in-situ qubit calibration dur-
232 D. M. Berns, M. S. Rudner, S. O. Valenzuela, K. K. Berggren, ing repetitive error detection,” Physical Review A 94, 032321
W. D. Oliver, L. S. Levitov, and T. P. Orlando, “Amplitude (2016).
spectroscopy of a solid-state artificial atom,” Nature 455, 51 254 S. J. Devitt, W. J. Munro, and K. Nemoto, “Quantum error

(2008). correction for beginners,” Reports on Progress in Physics 76,

233 M. S. Rudner, A. V. Shytov, L. S. Levitov, D. M. Berns, W. D. 076001 (2013).
Oliver, S. O. Valenzuela, and T. P. Orlando, “Quantum phase 255 D. Gottesman, “An Introduction to Quantum Error Cor-

tomography of a strongly driven qubit,” Phys. Rev. Lett. 101, rection and Fault-Tolerant Quantum Computation,” (2009),
190502 (2008). arXiv:0904.2557.
234 J. Q. You, Y.-x. Liu, and F. Nori, “Simultaneous cooling of 256 D. A. Lidar and T. A. Brun, eds., Quantum error correction

an artificial atom and its neighboring quantum system,” Phys. (Cambridge University Press, 2013) p. 666.
Rev. Lett. 100, 047001 (2008). 257 B. M. Terhal, “Quantum error correction for quantum memo-
235 M. Grajcar, S. Ashhab, J. R. Johansson, and F. Nori, “Lower ries,” Reviews of Modern Physics 87, 307–346 (2015).
limit on the achievable temperature in resonator-based sideband 258 D. Ristè, S. Poletto, M.-Z. Huang, A. Bruno, V. Vesterinen, O.-

cooling,” Phys. Rev. B 78, 035406 (2008). P. Saira, and L. DiCarlo, “Detecting bit-flip errors in a logical
236 J. Bylander, M. S. Rudner, A. V. Shytov, S. O. Valenzuela, qubit using stabilizer measurements,” Nature Communications
D. M. Berns, K. K. Berggren, L. S. Levitov, and W. D. Oliver, 6, 6983 (2015).
“Pulse imaging and nonadiabatic control of solid-state artificial 259 O.-P. Saira, J. Groen, J. Cramer, M. Meretska, G. de Lange,

atoms,” Phys. Rev. B 80, 220506 (2009). and L. DiCarlo, “Entanglement Genesis by Ancilla-Based Par-
237 W. D. Oliver and S. O. Valenzuela, “Large-amplitude driving of ity Measurement in 2D Circuit QED,” Physical Review Letters
a superconducting artificial atom,” Quantum Information Pro- 112, 070502 (2014).
cessing 8, 261–281 (2009). 260 R. Barends, L. Lamata, J. Kelly, L. Garcı́a-Álvarez, A. G.
238 S. Shevchenko, S. Ashhab, and F. Nori, “Landau–zener– Fowler, A. Megrant, E. Jeffrey, T. C. White, D. Sank, J. Y. Mu-
stückelberg interferometry,” Physics Reports 492, 1 – 30 (2010). tus, B. Campbell, Y. Chen, Z. Chen, B. Chiaro, A. Dunsworth,
239 S. Ashhab, P. C. de Groot, and F. Nori, “Speed limits for
I.-C. Hoi, C. Neill, P. J. J. O’Malley, C. Quintana, P. Roushan,
quantum gates in multiqubit systems,” Phys. Rev. A 85, 052327 A. Vainsencher, J. Wenner, E. Solano, and J. M. Martinis,
(2012). “Digital quantum simulation of fermionic models with a super-
240 M. G. Bason, M. Viteau, N. Malossi, P. Huillery, E. Arimondo,
conducting circuit,” Nature Communications 6, 7654 (2015).
D. Ciampini, R. Fazio, V. Giovannetti, R. Mannella, and 261 P. Jordan and E. Wigner, “Uber das Paulische Aquivalenzver-
O. Morsch, “High-fidelity quantum driving,” Nature Physics 8, bot,” Zeitschrift fur Physik 47, 631–651 (1928).
147–152 (2012). 262 U. Las Heras, L. Garcı́a-Álvarez, A. Mezzacapo, E. Solano, and
241 J. M. Martinis and M. R. Geller, “Fast adiabatic qubit gates
L. Lamata, “Fermionic models with superconducting circuits,”
using only σ z control,” Physical Review A 90, 022307 (2014). EPJ Quantum Technology 2, 8 (2015).
242 W. Press, W. Teukolsky, T. Vetterling, and B. Flannery, Nu- 263 P. Jordan and E. Wigner, “Uber das Paulische Aquivalenzver-
merical Recipes 3rd Edition: The Art of Scientific Comput- bot,” Zeitschrift fur Physik 47, 631–651 (1928).
ing, 3rd ed. (Cambridge University Press, New York, NY, USA, 264 V. Havlicek, A. D. Córcoles, K. Temme, A. W. Harrow,
2007). A. Kandala, J. M. Chow, and J. M. Gambetta, “Super-
243 J. Benhelm, G. Kirchmair, C. F. Roos, and R. Blatt, “Towards
vised learning with quantum enhanced feature spaces,” (2018),
fault-tolerant quantum computing with trapped ions,” Nature arXiv:1804.11326.
Physics 4, 463–466 (2008). 265 I. Buluta and F. Nori, “Quantum simulators,” Science 326, 108–
244 C. A. Ryan, M. Laforest, and R. Laflamme, “Randomized 111 (2009).
benchmarking of single- and multi-qubit control in liquid-state 266 I. M. Georgescu, S. Ashhab, and F. Nori, “Quantum simula-
NMR quantum information processing,” New Journal of Physics tion,” Rev. Mod. Phys. 86, 153–185 (2014).
11, 013034 (2009). 267 P. O’Malley, R. Babbush, I. Kivlichan, J. Romero, J. McClean,
245 E. Magesan, J. M. Gambetta, and J. Emerson, “Scalable R. Barends, J. Kelly, P. Roushan, A. Tranter, N. Ding, B. Camp-
and Robust Randomized Benchmarking of Quantum Processes,” bell, Y. Chen, Z. Chen, B. Chiaro, A. Dunsworth, A. Fowler,
Physical Review Letters 106, 180504 (2011). E. Jeffrey, E. Lucero, A. Megrant, J. Mutus, M. Neeley, C. Neill,
246 A. D. Córcoles, J. M. Gambetta, J. M. Chow, J. A. Smolin,
C. Quintana, D. Sank, A. Vainsencher, J. Wenner, T. White,
M. Ware, J. Strand, B. L. T. Plourde, and M. Steffen, “Process P. Coveney, P. Love, H. Neven, A. Aspuru-Guzik, and J. Mar-
verification of two-qubit quantum gates by randomized bench- tinis, “Scalable Quantum Simulation of Molecular Energies,”
marking,” Physical Review A 87, 030301 (2013). Physical Review X 6, 031007 (2016).
247 A. G. Fowler, M. Mariantoni, J. M. Martinis, and A. N. Cleland, 268 M. Baur, A. Fedorov, L. Steffen, S. Filipp, M. P. da Silva, and
“Surface codes: Towards practical large-scale quantum compu- A. Wallraff, “Benchmarking a Quantum Teleportation Protocol
tation,” Physical Review A 86, 032324 (2012). in Superconducting Circuits Using Tomography and an Entan-
248 Y. Tomita and K. M. Svore, “Low-distance surface codes under
glement Witness,” Physical Review Letters 108, 040502 (2012).
realistic quantum noise,” Physical Review A 90, 062320 (2014). 269 L. Steffen, Y. Salathe, M. Oppliger, P. Kurpiers, M. Baur,
249 T. E. O’Brien, B. Tarasinski, and L. DiCarlo, “Density-matrix
C. Lang, C. Eichler, G. Puebla-Hellmann, A. Fedorov, and
simulation of small surface codes under current and projected A. Wallraff, “Deterministic quantum teleportation with feed-
experimental noise,” npj Quantum Information 3, 39 (2017). forward in a solid state system,” Nature 500, 319–322 (2013).
250 N. H. Nickerson, “Error correcting power of small topological 270 C. Neill, P. Roushan, K. Kechedzhi, S. Boixo, S. V. Isakov,
codes,” (2016), arXiv:1609.01753. V. Smelyanskiy, A. Megrant, B. Chiaro, A. Dunsworth, K. Arya,
251 E. T. Campbell, B. M. Terhal, and C. Vuillot, “Roads to- R. Barends, B. Burkett, Y. Chen, Z. Chen, A. Fowler, B. Foxen,
wards fault-tolerant universal quantum computation,” Nature M. Giustina, R. Graff, E. Jeffrey, T. Huang, J. Kelly, P. Klimov,
549, 172–179 (2017). E. Lucero, J. Mutus, M. Neeley, C. Quintana, D. Sank,
252 S. Benjamin and J. Kelly, “Solving a wonderful problem,” Na-
A. Vainsencher, J. Wenner, T. C. White, H. Neven, and J. M.
ture Materials 14, 561–563 (2015). Martinis, “A blueprint for demonstrating quantum supremacy
253 J. Kelly, R. Barends, A. G. Fowler, A. Megrant, E. Jeffrey, T. C.
with superconducting qubits.” Science (New York, N.Y.) 360,
White, D. Sank, J. Y. Mutus, B. Campbell, Y. Chen, Z. Chen, 195–199 (2018).
B. Chiaro, A. Dunsworth, E. Lucero, M. Neeley, C. Neill, P. J. J. 271 A. W. Harrow, A. Hassidim, and S. Lloyd, “Quantum Algo-
O’Malley, C. Quintana, P. Roushan, A. Vainsencher, J. Wen- rithm for Linear Systems of Equations,” Physical Review Let-
 63

ters 103, 150502 (2009). J. M. Chow, and J. M. Gambetta, “Hardware-efficient vari-

272 Y. Zheng, C. Song, M.-C. Chen, B. Xia, W. Liu, Q. Guo, ational quantum eigensolver for small molecules and quantum
L. Zhang, D. Xu, H. Deng, K. Huang, Y. Wu, Z. Yan, D. Zheng, magnets,” Nature 549, 242–246 (2017).
L. Lu, J.-W. Pan, H. Wang, C.-Y. Lu, and X. Zhu, “Solving 289 S. Bravyi, J. M. Gambetta, A. Mezzacapo, and K. Temme, “Ta-

Systems of Linear Equations with a Superconducting Quantum pering off qubits to simulate fermionic Hamiltonians,” (2017),
Processor,” Physical Review Letters 118, 210504 (2017). arXiv:1701.08213.
273 T. Larsen, K. Petersson, F. Kuemmeth, T. Jespersen, 290 “IBM Quantum Experience,” https://quantumexperience.ng.bluemix.net/qx/

P. Krogstrup, J. Nygård, and C. Marcus, “Semiconductor- 291 M. Takita, A. W. Cross, A. Córcoles, J. M. Chow, and J. M.

Nanowire-Based Superconducting Qubit,” Physical Review Let- Gambetta, “Experimental Demonstration of Fault-Tolerant
ters 115, 127001 (2015). State Preparation with Superconducting Qubits,” Physical Re-
274 G. de Lange, B. van Heck, A. Bruno, D. van Woerkom, view Letters 119, 180501 (2017).
A. Geresdi, S. Plissard, E. Bakkers, A. Akhmerov, and L. Di- 292 D. W. Leung, M. A. Nielsen, I. L. Chuang, and Y. Ya-
Carlo, “Realization of Microwave Quantum Circuits Using Hy- mamoto, “Approximate quantum error correction can lead to
brid Superconducting-Semiconducting Nanowire Josephson El- better codes,” Physical Review A 56, 2567–2573 (1997).
ements,” Physical Review Letters 115, 127002 (2015). 293 D. Gottesman, “Quantum fault tolerance in small experiments,”
275 J. I.-J. Wang, D. Rodan-Legrain, L. Bretheau, D. L. Campbell, (2016), arXiv:1610.03507.
B. Kannan, D. Kim, M. Kjaergaard, P. Krantz, G. O. Samach, 294 C. Vuillot, “Is error detection helpful on IBM 5Q chips ?”

F. Yan, J. L. Yoder, K. Watanabe, T. Taniguchi, T. P. Orlando, (2017), arXiv:1705.08957.

S. Gustavsson, P. Jarillo-Herrero, and W. D. Oliver, “Coherent 295 D. Ristè, M. P. da Silva, C. A. Ryan, A. W. Cross, A. D.

control of a hybrid superconducting circuit made with graphene- Córcoles, J. A. Smolin, J. M. Gambetta, J. M. Chow, and
based van der Waals heterostructures,” Nature Nanotechnology B. R. Johnson, “Demonstration of quantum advantage in ma-
14, 120–125 (2019). chine learning,” npj Quantum Information 3, 16 (2017).
276 L. Casparis, T. Larsen, M. Olsen, F. Kuemmeth, P. Krogstrup, 296 D. J. C. MacKay, Information theory, inference, and learning

J. Nygård, K. Petersson, and C. Marcus, “Gatemon Bench- algorithms (Cambridge University Press, 2003) p. 628.
marking and Two-Qubit Operations,” Physical Review Letters 297 P. Rebentrost, M. Mohseni, and S. Lloyd, “Quantum Support

116, 150505 (2016). Vector Machine for Big Data Classification,” Physical Review
277 G. S. Paraoanu, “Microwave-induced coupling of superconduct- Letters 113, 130503 (2014).
ing qubits,” Physical Review B 74, 140504 (2006). 298 J. Biamonte, P. Wittek, N. Pancotti, P. Rebentrost, N. Wiebe,
278 C. Rigetti and M. Devoret, “Fully microwave-tunable universal and S. Lloyd, “Quantum machine learning,” Nature 549, 195–
gates in superconducting qubits with linear couplings and fixed 202 (2017).
transition frequencies,” Physical Review B 81, 134507 (2010). 299 S. Poletto, J. M. Gambetta, S. T. Merkel, J. A. Smolin, J. M.
279 P. C. de Groot, J. Lisenfeld, R. N. Schouten, S. Ashhab, Chow, A. D. Córcoles, G. A. Keefe, M. B. Rothwell, J. R.
A. Lupaşcu, C. J. P. M. Harmans, and J. E. Mooij, “Se- Rozen, D. W. Abraham, C. Rigetti, and M. Steffen, “Entan-
lective darkening of degenerate transitions demonstrated with glement of Two Superconducting Qubits in a Waveguide Cavity
two superconducting quantum bits,” Nature Physics 6, 763–766 via Monochromatic Two-Photon Excitation,” Physical Review
(2010). Letters 109, 240505 (2012).
280 J. M. Gambetta, “Control of Superconducting Qubits,” in Lec- 300 K. Mølmer and A. Sørensen, “Multiparticle Entanglement of

ture Notes of the 44th IFF Spring School, edited by D. P. Di- Hot Trapped Ions,” Physical Review Letters 82, 1835–1838
Vincenzo (2013). (1999).
281 J. M. Chow, J. M. Gambetta, A. D. Córcoles, S. T. Merkel, 301 J. Colless, V. Ramasesh, D. Dahlen, M. Blok, M. Kimchi-

J. A. Smolin, C. Rigetti, S. Poletto, G. A. Keefe, M. B. Roth- Schwartz, J. McClean, J. Carter, W. de Jong, and I. Sid-
well, J. R. Rozen, M. B. Ketchen, and M. Steffen, “Universal diqi, “Computation of Molecular Spectra on a Quantum Pro-
Quantum Gate Set Approaching Fault-Tolerant Thresholds with cessor with an Error-Resilient Algorithm,” Physical Review X
Superconducting Qubits,” Physical Review Letters 109, 060501 8, 011021 (2018).
(2012). 302 A. W. Cross and J. M. Gambetta, “Optimized pulse shapes for
282 J. M. Chow, A. D. Córcoles, J. M. Gambetta, C. Rigetti, B. R. a resonator-induced phase gate,” Physical Review A 91, 032325
Johnson, J. A. Smolin, J. R. Rozen, G. A. Keefe, M. B. Roth- (2015).
well, M. B. Ketchen, and M. Steffen, “Simple All-Microwave 303 S. Puri and A. Blais, “High-Fidelity Resonator-Induced Phase

Entangling Gate for Fixed-Frequency Superconducting Qubits,” Gate with Single-Mode Squeezing,” Physical Review Letters
Physical Review Letters 107, 080502 (2011). 116, 180501 (2016).
283 J. M. Chow, J. M. Gambetta, E. Magesan, D. W. Abraham, 304 H. Paik, A. Mezzacapo, M. Sandberg, D. McClure, B. Abdo,

A. W. Cross, B. R. Johnson, N. A. Masluk, C. A. Ryan, J. A. A. Córcoles, O. Dial, D. Bogorin, B. Plourde, M. Steffen,

Smolin, S. J. Srinivasan, and M. Steffen, “Implementing a A. Cross, J. Gambetta, and J. M. Chow, “Experimental Demon-
strand of a scalable fault-tolerant quantum computing fabric,” stration of a Resonator-Induced Phase Gate in a Multiqubit
Nature Communications 5, 4015 (2014). Circuit-QED System,” Physical Review Letters 117, 250502
284 A. Córcoles, E. Magesan, S. J. Srinivasan, A. W. Cross, M. Stef- (2016).
fen, J. M. Gambetta, and J. M. Chow, “Demonstration of a 305 P. Bertet, C. J. P. M. Harmans, and J. E. Mooij, “Parametric

quantum error detection code using a square lattice of four su- coupling for superconducting qubits,” Physical Review B 73,
perconducting qubits,” Nature Communications 6, 6979 (2015). 064512 (2006).
285 M. Takita, A. Córcoles, E. Magesan, B. Abdo, M. Brink, 306 F. Wulschner, J. Goetz, F. R. Koessel, E. Hoffmann, A. Baust,

A. Cross, J. M. Chow, and J. M. Gambetta, “Demonstration P. Eder, M. Fischer, M. Haeberlein, M. J. Schwarz, M. Pern-
of Weight-Four Parity Measurements in the Surface Code Ar- peintner, E. Xie, L. Zhong, C. W. Zollitsch, B. Peropadre, J.-J.
chitecture,” Physical Review Letters 117, 210505 (2016). Garcia Ripoll, E. Solano, K. G. Fedorov, E. P. Menzel, F. Deppe,
286 E. Magesan and J. M. Gambetta, “Effective Hamiltonian models A. Marx, and R. Gross, “Tunable coupling of transmission-line
of the cross-resonance gate,” (2018), arXiv:1804.04073. microwave resonators mediated by an rf SQUID,” EPJ Quan-
287 S. Kirchhoff, T. Keßler, P. J. Liebermann, E. Assémat, tum Technology 3, 10 (2016).
S. Machnes, F. Motzoi, and F. K. Wilhelm, “Optimized cross- 307 M. Allman, J. Whittaker, M. Castellanos-Beltran, K. Cicak,

resonance gate for coupled transmon systems,” Physical Review F. da Silva, M. DeFeo, F. Lecocq, A. Sirois, J. Teufel, J. Au-
A 97, 042348 (2018). mentado, and R. Simmonds, “Tunable Resonant and Nonreso-
288 A. Kandala, A. Mezzacapo, K. Temme, M. Takita, M. Brink, nant Interactions between a Phase Qubit and L C Resonator,”
 64

Physical Review Letters 112, 123601 (2014). 326 F. Marki and C. Marki, “Mixer Basics Primer - A Tutorial for
308 M. Wallquist, V. S. Shumeiko, and G. Wendin, “Selective cou- RF & Microwave Mixers,” (2010).
pling of superconducting charge qubits mediated by a tunable 327 L. Mandel and E. Wolf, Optical Coherence and Quantum Optics

stripline cavity,” Physical Review B 74, 224506 (2006). (Cambridge University Press, 1995).
309 K. Harrabi, F. Yoshihara, A. O. Niskanen, Y. Nakamura, and 328 C. M. Caves, “Quantum limits on noise in linear amplifiers,”

J. S. Tsai, “Engineered selection rules for tunable coupling in Phys. Rev. D 26, 1817–1839 (1982).
a superconducting quantum circuit,” Physical Review B 79, 329 M. Simoen, C. W. S. Chang, P. Krantz, J. Bylander, W. Wust-

020507 (2009). mann, V. Shumeiko, and P. Delsing, “Characterization of a

310 H. Wang, M. Mariantoni, R. C. Bialczak, M. Lenander, multimode coplanar waveguide parametric amplifier,” Journal
E. Lucero, M. Neeley, A. D. O’Connell, D. Sank, M. Weides, of Applied Physics 118, 154501 (2015).
J. Wenner, T. Yamamoto, Y. Yin, J. Zhao, J. M. Martinis, and 330 C. Macklin, K. O’Brien, D. Hover, M. E. Schwartz,
A. N. Cleland, “Deterministic Entanglement of Photons in Two V. Bolkhovsky, X. Zhang, W. D. Oliver, and I. Siddiqi, “A
Superconducting Microwave Resonators,” Physical Review Let- near–quantum-limited josephson traveling-wave parametric am-
ters 106, 060401 (2011). plifier,” 350, 307–310 (2015).
311 J. D. Whittaker, F. C. S. da Silva, M. S. Allman, F. Lecocq, 331 E. M. Purcell, H. C. Torrey, and R. V. Pound, “Resonance

K. Cicak, A. J. Sirois, J. D. Teufel, J. Aumentado, and R. W. absorption by nuclear magnetic moments in a solid,” Phys. Rev.
Simmonds, “Tunable-cavity QED with phase qubits,” Physical 69, 37–38 (1946).
Review B 90, 024513 (2014). 332 A. A. Houck, J. A. Schreier, B. R. Johnson, J. M. Chow, J. Koch,
312 C. K. Andersen and K. Mølmer, “Multifrequency modes in J. M. Gambetta, D. I. Schuster, L. Frunzio, M. H. Devoret,
superconducting resonators: Bridging frequency gaps in off- S. M. Girvin, and R. J. Schoelkopf, “Controlling the sponta-
resonant couplings,” Physical Review A 91, 023828 (2015). neous emission of a superconducting transmon qubit,” Phys.
313 J. S. Otterbach, R. Manenti, N. Alidoust, A. Bestwick, Rev. Lett. 101, 080502 (2008).
M. Block, B. Bloom, S. Caldwell, N. Didier, E. S. Fried, 333 M. D. Reed, B. R. Johnson, A. A. Houck, L. DiCarlo, J. M.

S. Hong, P. Karalekas, C. B. Osborn, A. Papageorge, E. C. Chow, D. I. Schuster, L. Frunzio, and R. J. Schoelkopf, “Fast
Peterson, G. Prawiroatmodjo, N. Rubin, C. A. Ryan, D. Scara- reset and suppressing spontaneous emission of a superconduct-
belli, M. Scheer, E. A. Sete, P. Sivarajah, R. S. Smith, A. Staley, ing qubit,” Applied Physics Letters 96, 203110 (2010).
N. Tezak, W. J. Zeng, A. Hudson, B. R. Johnson, M. Reagor, 334 E. Jeffrey, D. Sank, J. Y. Mutus, T. C. White, J. Kelly,

M. P. da Silva, and C. Rigetti, “Unsupervised Machine Learning R. Barends, Y. Chen, Z. Chen, B. Chiaro, a. Dunsworth,
on a Hybrid Quantum Computer,” (2017), arXiv:1712.05771. a. Megrant, P. J. O’Malley, C. Neill, P. Roushan, a. Vainsencher,
314 V. B. Braginsky and F. Y. Khalili, “Quantum nondemolition J. Wenner, a. N. Cleland, and J. M. Martinis, “Fast accurate
measurements: The route from toys to tools,” Reviews of Mod- state measurement with superconducting qubits,” Physical Re-
ern Physics 68, 1–11 (1996). view Letters 112, 1–5 (2014), arXiv:1401.0257.
315 E. T. Jaynes and F. W. Cummings, “Comparison of quan- 335 N. T. Bronn, Y. Liu, J. B. Hertzberg, A. D. Córcoles, A. A.

tum and semiclassical radiation theories with application to the Houck, J. M. Gambetta, and J. M. Chow, “Broadband filters
beam maser,” Proc. IEEE 51, 89–109 (2004). for abatement of spontaneous emission in circuit quantum elec-
316 B. W. Shore and P. L. Knight, “The jaynes-cummings model,” trodynamics,” Applied Physics Letters 107, 172601 (2015).
Journal of Modern Optics 40, 1195–1238 (1993). 336 I.-C. Hoi, C. M. Wilson, G. Johansson, T. Palomaki, B. Per-
317 C. C. Gerry and P. L. Knight, Introductory Quantum Optics opadre, and P. Delsing, “Demonstration of a single-photon
(Cambridge University Press, Cambridge, England, 2005). router in the microwave regime,” Phys. Rev. Lett. 107, 073601
318 M. Boissonneault, J. M. Gambetta, and A. Blais, “Dispersive (2011).
regime of circuit qed: Photon-dependent qubit dephasing and 337 K. Inomata, Z. Lin, K. Koshino, W. D. Oliver, J.-S. Tsai, T. Ya-

relaxation rates,” Phys. Rev. A 79, 013819 (2009). mamoto, and Y. Nakamura, “Single microwave-photon detector
319 J. Gambetta, A. Blais, D. I. Schuster, A. Wallraff, L. Frunzio, using an artificial Λ-type three-level system,” Nature Commu-
J. Majer, M. H. Devoret, S. M. Girvin, and R. J. Schoelkopf, nications 7, 12303 (2016).
“Qubit-photon interactions in a cavity: Measurement-induced 338 D. M. Pozar, Microwave Engineering (John Wiley & Sons,

dephasing and number splitting,” Phys. Rev. A 74, 042318 2004).

(2006). 339 H. A. Haus and J. A. Mullen, “Quantum noise in linear ampli-
320 D. Sank, Z. Chen, M. Khezri, J. Kelly, R. Barends, B. Camp- fiers,” Phys. Rev. 128, 2407–2413 (1962).
bell, Y. Chen, B. Chiaro, A. Dunsworth, A. Fowler, E. Jeffrey, 340 E. Flurin, The Josephson Mixer, a Swiss army knife for mi-

E. Lucero, A. Megrant, J. Mutus, M. Neeley, C. Neill, P. J. J. crowave quantum optics, Ph.D. thesis (2014).
O’Malley, C. Quintana, P. Roushan, A. Vainsencher, T. White, 341 K. Shimoda, H. Takahasi, and C. H. Townes, “Fluctuations in

J. Wenner, A. N. Korotkov, and J. M. Martinis, “Measurement- amplification of quanta with application to maser amplifiers,”
induced state transitions in a superconducting qubit: Beyond Journal of the Physical Society of Japan 12, 686–700 (1957).
the rotating wave approximation,” Phys. Rev. Lett. 117, 190503 342 A. A. Clerk, S. M. Girvin, and A. D. Stone, “Quantum-limited

(2016). measurement and information in mesoscopic detectors,” Phys.

321 M. D. Reed, L. DiCarlo, B. R. Johnson, L. Sun, D. I. Schus- Rev. B 67, 165324 (2003).
ter, L. Frunzio, and R. J. Schoelkopf, “High-fidelity readout 343 C. C. Bultink, B. Tarasinski, N. Haandbæk, S. Poletto,

in circuit quantum electrodynamics using the jaynes-cummings N. Haider, D. J. Michalak, A. Bruno, and L. DiCarlo, “General
nonlinearity,” Phys. Rev. Lett. 105, 173601 (2010). method for extracting the quantum efficiency of dispersive qubit
322 A. J. Kerman, “Quantum information processing using quasi- readout in circuit qed,” Applied Physics Letters 112, 092601
classical electromagnetic interactions between qubits and elec- (2018).
trical resonators,” New Journal of Physics 15, 123011 (2013). 344 B. Yurke, P. G. Kaminsky, R. E. Miller, E. a. Whittaker, a. D.
323 S. Richer and D. DiVincenzo, “Circuit design implementing Smith, a. H. Silver, and R. W. Simon, “Observation of 4.2-K
longitudinal coupling: A scalable scheme for superconducting equilibrium-noise squeezing via a Josephson-parametric ampli-
qubits,” Phys. Rev. B 93, 134501 (2016). fier,” Physical Review Letters 60, 764–767 (1988).
324 G. Zhang, Y. Liu, J. J. Raftery, and A. A. Houck, “Suppression 345 C. W. Gardiner, “Inhibition of atomic phase decays by squeezed

of photon shot noise dephasing in a tunable coupling supercon- light: A direct effect of squeezing,” Phys. Rev. Lett. 56, 1917–
ducting qubit,” npj Quantum Information 3, 1 (2017). 1920 (1986).
325 M. Reed, Entanglement and Quantum Error Correction with 346 K. W. Murch, S. J. Weber, K. M. Beck, E. Ginossar, and

Superconducting Qubits, Ph.D. thesis (2013). I. Siddiqi, “Reduction of the radiative decay of atomic coherence
 65

in squeezed vacuum,” Nature 499, 62 (2013). 366 J. C. Lee, W. D. Oliver, K. K. Berggren, and T. P. Orlando,
347 D. M. Toyli, A. W. Eddins, S. Boutin, S. Puri, D. Hover, “Nonlinear resonant behavior of a dispersive readout circuit for
V. Bolkhovsky, W. D. Oliver, A. Blais, and I. Siddiqi, “Reso- a superconducting flux qubit,” Phys. Rev. B 75, 144505 (2007).
nance fluorescence from an artificial atom in squeezed vacuum,” 367 F. Mallet, F. R. Ong, A. Palacios-Laloy, F. Nguyen, P. Bertet,

Phys. Rev. X 6, 031004 (2016). D. Vion, and D. Esteve, “Single-shot qubit readout in circuit
348 C. Eichler, D. Bozyigit, C. Lang, M. Baur, L. Steffen, J. M. quantum electrodynamics,” Nature Physics 5, 791–795 (2009).
Fink, S. Filipp, and A. Wallraff, “Observation of two-mode 368 P. Krantz, A. Bengtsson, M. Simoen, S. Gustavsson,
squeezing in the microwave frequency domain,” Phys. Rev. Lett. V. Shumeiko, W. D. Oliver, C. M. Wilson, P. Delsing, and
107, 113601 (2011). J. Bylander, “Single-shot Readout of a Superconducting Qubit
349 A. Roy and M. Devoret, “Introduction to parametric amplifi- using a Josephson Parametric Oscillator,” Nature Communica-
cation of quantum signals with josephson circuits,” Comptes tions 7, 11417 (2016), 1508.02886.
Rendus Physique 17, 740 – 755 (2016), quantum microwaves / 369 P. Krantz, The Josephson parametric oscillator - From micro-

Micro-ondes quantiques. scopic studies to single-shot qubit readout, Ph.D. thesis (2016).
350 W. Wustmann and V. Shumeiko, “Parametric resonance in tun- 370 Z. R. Lin, K. Inomata, W. D. Oliver, K. Koshino, Y. Nakamura,

able superconducting cavities,” Physical Review B - Condensed J. S. Tsai, and T. Yamamoto, “Single-shot readout of a super-
Matter and Materials Physics 87, 1–23 (2013), arXiv:1302.3484. conducting flux qubit with a flux-driven josephson parametric
351 G. J. Dolan, “Offset masks for lift-off photoprocessing,” Applied amplifier,” Applied Physics Letters 103, 132602 (2013).
Physics Letters 31, 337–339 (1977). 371 Z. R. Lin, K. Inomata, K. Koshino, W. D. Oliver, Y. Naka-
352 B. Yurke, L. R. Corruccini, P. G. Kaminsky, L. W. Rupp, A. D. mura, J. S. Tsai, and T. Yamamoto, “Josephson parametric
Smith, A. H. Silver, R. W. Simon, and E. A. Whittaker, “Ob- phase-locked oscillator and its application to dispersive read-
servation of parametric amplification and deamplification in a out of superconducting qubits,” Nature Communications 5, 15
josephson parametric amplifier,” Phys. Rev. A 39, 2519–2533 (2014), arXiv:1407.7195.
(1989). 372 J. R. Johansson, G. Johansson, C. M. Wilson, and F. Nori,
353 W. Wustmann and V. Shumeiko, “Nondegenerate parametric “Dynamical casimir effect in a superconducting coplanar waveg-
resonance in a tunable superconducting cavity,” Phys. Rev. Ap- uide,” Phys. Rev. Lett. 103, 147003 (2009).
plied 8, 024018 (2017). 373 J. R. Johansson, G. Johansson, C. M. Wilson, and F. Nori, “Dy-
354 S. Wahlsten, S. Rudner, and T. Claeson, “Arrays of Josephson namical casimir effect in superconducting microwave circuits,”
tunnel junctions as parametric amplifiers,” Journal of Applied Phys. Rev. A 82, 052509 (2010).
Physics 49 (1978), 10.1063/1.325341. 374 C. M. Wilson, G. Johansson, A. Pourkabirian, M. Simoen, J. R.
355 H. K. Olsson and T. Claeson, “Low-noise Josephson parametric Johansson, T. Duty, F. Nori, and P. Delsing, “Observation
amplification and oscillations at 9 GHz,” Journal of Applied of the dynamical Casimir effect in a superconducting circuit,”
Physics 64, 5234–5243 (1988). Nature 479, 376 (2011).
356 I. Siddiqi, R. Vijay, F. Pierre, C. M. Wilson, M. Metcalfe, 375 P. Krantz, Y. Reshitnyk, W. Wustmann, J. Bylander, S. Gus-

C. Rigetti, L. Frunzio, and M. H. Devoret, “Rf-driven joseph- tavsson, W. D. Oliver, T. Duty, V. Shumeiko, and P. Delsing,
son bifurcation amplifier for quantum measurement,” Phys. Rev. “Investigation of nonlinear effects in Josephson parametric oscil-
Lett. 93, 207002 (2004). lators used in circuit quantum electrodynamics,” New Journal
357 E. A. Tholén, A. Ergül, E. M. Doherty, F. M. Weber, F. Grégis, of Physics 15, 105002 (2013).
and D. B. Haviland, “Nonlinearities and parametric amplifica- 376 P. Krantz, Parametrically pumped superconducting circuits

tion in superconducting coplanar waveguide resonators,” Ap- (Chalmers University of Technology, 2013).
plied Physics Letters 90, 253509 (2007), arXiv:0702280. 377 I.-M. Svensson, A. Bengtsson, P. Krantz, J. Bylander,
358 M. A. Castellanos-Beltran and K. W. Lehnert, “Widely tunable V. Shumeiko, and P. Delsing, “Period-tripling subharmonic os-
parametric amplifier based on a superconducting quantum in- cillations in a driven superconducting resonator,” Phys. Rev. B
terference device array resonator,” Applied Physics Letters 91 96, 174503 (2017).
(2007), 10.1063/1.2773988, arXiv:0706.2373. 378 I.-M. Svensson, M. Pierre, M. Simoen, W. Wustmann,
359 T. Yamamoto, K. Inomata, M. Watanabe, K. Matsuba, P. Krantz, A. Bengtsson, G. Johansson, J. Bylander,
T. Miyazaki, W. D. Oliver, Y. Nakamura, and J. S. Tsai, “Flux- V. Shumeiko, and P. Delsing, “Microwave photon generation
driven josephson parametric amplifier,” Applied Physics Letters in a doubly tunable superconducting resonator,” Journal of
93, 042510 (2008). Physics: Conference Series 969, 012146 (2018).
360 M. A. Castellanos-Beltran, K. D. Irwin, G. C. Hilton, L. R. Vale, 379 A. Bengtsson, P. Krantz, M. Simoen, I.-M. Svensson, B. Schnei-

and K. W. Lehnert, “Amplification and squeezing of quantum der, V. Shumeiko, P. Delsing, and J. Bylander, “Nondegenerate
noise with a tunable Josephson metamaterial,” Nature Physics parametric oscillations in a tunable superconducting resonator,”
4, 929 (2008). Phys. Rev. B 97, 144502 (2018).
361 C. Wilson, T. Duty, M. Sandberg, F. Persson, V. Shumeiko, and 380 M. Simoen, Parametric interactions with signals and the vac-

P. Delsing, “Photon Generation in an Electromagnetic Cavity uum, Ph.D. thesis (2015).

with a Time-Dependent Boundary,” Physical Review Letters 381 C. Eichler, Y. Salathe, J. Mlynek, S. Schmidt, and A. Wallraff,

105, 233907 (2010). “Quantum-limited amplification and entanglement in coupled

362 K. M. Sundqvist, S. Kintas, M. Simoen, P. Krantz, M. Sandberg, nonlinear resonators,” Phys. Rev. Lett. 113, 110502 (2014).
C. M. Wilson, and P. Delsing, “The pumpistor: A linearized 382 N. Bergeal, R. Vijay, V. E. Manucharyan, I. Siddiqi, R. J.

model of a flux-pumped superconducting quantum interference Schoelkopf, S. M. Girvin, and M. H. Devoret, “Analog infor-
device for use as a negative-resistance parametric amplifier,” mation processing at the quantum limit with a Josephson ring
Applied Physics Letters 103, 102603 (2013). modulator,” Nature Physics 6, 296 (2010).
363 K. M. Sundqvist and P. Delsing, “Negative-resistance models for 383 N. Bergeal, F. Schackert, M. Metcalfe, R. Vijay, V. E.

parametrically flux-pumped superconducting quantum interfer- Manucharyan, L. Frunzio, D. E. Prober, R. J. Schoelkopf, S. M.

ence devices,” EPJ Quantum Technology 1, 6 (2014). Girvin, and M. H. Devoret, “Phase-preserving amplification
364 M. I. Dykman, C. M. Maloney, V. N. Smelyanskiy, and M. Sil- near the quantum limit with a Josephson ring modulator,” Na-
verstein, “Fluctuational phase-flip transitions in parametrically ture 465, 64 (2010).
driven oscillators,” Physical Review E 57, 5202–5212 (1998). 384 B. Abdo, F. Schackert, M. Hatridge, C. Rigetti, and M. De-
365 R. Vijay, M. H. Devoret, and I. Siddiqi, “Invited review arti- voret, “Josephson amplifier for qubit readout,” Applied Physics
cle: The josephson bifurcation amplifier,” Review of Scientific Letters 99, 162506 (2011).
Instruments 80, 111101 (2009). 385 B. Abdo, A. Kamal, and M. Devoret, “Nondegenerate three-
 66

wave mixing with the josephson ring modulator,” Phys. Rev. B Physical Society of Japan 88, 061007 (2019).
87, 014508 (2013). 403 M. Reagor, H. Paik, G. Catelani, L. Sun, C. Axline, E. Holland,
386 G. Liu, T.-C. Chien, X. Cao, O. Lanes, E. Alpern, D. Pekker, I. M. Pop, N. A. Masluk, T. Brecht, L. Frunzio, M. H. Devoret,
and M. Hatridge, “Josephson parametric converter saturation L. Glazman, and R. J. Schoelkopf, “Reaching 10 ms single pho-
and higher order effects,” Applied Physics Letters 111, 202603 ton lifetimes for superconducting aluminum cavities,” Applied
(2017). Physics Letters 102, 192604 (2013).
387 Y. Chen, D. Sank, P. O’Malley, T. White, R. Barends, 404 W. Pfaff, C. J. Axline, L. D. Burkhart, U. Vool, P. Reinhold,

B. Chiaro, J. Kelly, E. Lucero, M. Mariantoni, A. Megrant, L. Frunzio, L. Jiang, M. H. Devoret, and R. J. Schoelkopf,
C. Neill, A. Vainsencher, J. Wenner, Y. Yin, A. N. Cleland, “Controlled release of multiphoton quantum states from a mi-
and J. M. Martinis, “Multiplexed dispersive readout of super- crowave cavity memory,” Nature Physics 13, 882 (2017).
conducting phase qubits,” Applied Physics Letters 101, 182601 405 R. W. Heeres, P. Reinhold, N. Ofek, L. Frunzio, L. Jiang, M. H.

(2012). Devoret, and R. J. Schoelkopf, “Implementing a universal gate

388 M. Jerger, S. Poletto, P. Macha, U. Hübner, E. Il’ichev, and set on a logical qubit encoded in an oscillator,” Nature Commu-
A. V. Ustinov, “Frequency division multiplexing readout and nications 8, 94 (2017).
simultaneous manipulation of an array of flux qubits,” Applied 406 S. Rosenblum, Y. Y. Gao, P. Reinhold, C. Wang, C. J. Axline,

Physics Letters 101, 042604 (2012). L. Frunzio, S. M. Girvin, L. Jiang, M. Mirrahimi, M. H. De-
389 B. J. Chapman, E. I. Rosenthal, J. Kerckhoff, L. R. Vale, G. C. voret, and R. J. Schoelkopf, “A CNOT gate between multipho-
Hilton, and K. W. Lehnert, “Single-sideband modulator for fre- ton qubits encoded in two cavities,” Nature Communications 9,
quency domain multiplexing of superconducting qubit readout,” 652 (2018).
Applied Physics Letters 110, 162601 (2017). 407 L. Hu, Y. Ma, W. Cai, X. Mu, Y. Xu, W. Wang, Y. Wu,
390 T. Roy, S. Kundu, M. Chand, A. M. Vadiraj, A. Ranadive, H. Wang, Y. P. Song, C.-L. Zou, S. M. Girvin, L.-M. Duan,
N. Nehra, M. P. Patankar, J. Aumentado, A. A. Clerk, and and L. Sun, “Quantum error correction and universal gate set
R. Vijay, “Broadband parametric amplification with impedance operation on a binomial bosonic logical qubit,” Nature Physics
engineering: Beyond the gain-bandwidth product,” Applied (2019), 10.1038/s41567-018-0414-3.
Physics Letters 107, 262601 (2015). 408 S. K. Tolpygo, V. Bolkhovsky, T. J. Weir, L. M. Johnson, M. A.
391 B. Ho Eom, P. K. Day, H. G. LeDuc, and J. Zmuidzinas, “A Gouker, and W. D. Oliver, “Fabrication process and proper-
wideband, low-noise superconducting amplifier with high dy- ties of fully-planarized deep-submicron nb/al–alox/nb joseph-
namic range,” Nature Physics 8, 623 (2012). son junctions for vlsi circuits,” IEEE Transactions on Applied
392 C. Bockstiegel, J. Gao, M. R. Vissers, M. Sandberg, S. Chaud- Superconductivity 25, 1–12 (2015).
huri, A. Sanders, L. R. Vale, K. Irwin, and D. P. Pappas, 409 J. H. Bejanin, T. G. McConkey, J. R. Rinehart, C. T. Earnest,

“Development of a Broadband NbTiN Traveling Wave Para- C. R. H. McRae, D. Shiri, J. D. Bateman, Y. Rohanizade-
metric Amplifier for MKID Readout,” J Low Temp Phys 176, gan, B. Penava, P. Breul, S. Royak, M. Zapatka, A. G. Fowler,
476 (2014). and M. Mariantoni, “Three-Dimensional Wiring for Extensible
393 A. A. Adamyan, S. E. de Graaf, S. E. Kubatkin, and Quantum Computing: The Quantum Socket,” Physical Review
A. V. Danilov, “Superconducting microwave parametric ampli- Applied 6, 044010 (2016).
fier based on a quasi-fractal slow propagation line,” Journal of 410 M. Vahidpour, W. O’Brien, J. T. Whyland, J. Angeles, J. Mar-

Applied Physics 119, 083901 (2016). shall, D. Scarabelli, G. Crossman, K. Yadav, Y. Mohan,
394 K. O’Brien, C. Macklin, I. Siddiqi, and X. Zhang, “Resonant C. Bui, V. Rawat, R. Renzas, N. Vodrahalli, A. Bestwick, and
phase matching of josephson junction traveling wave parametric C. Rigetti, “Superconducting Through-Silicon Vias for Quan-
amplifiers,” Phys. Rev. Lett. 113, 157001 (2014). tum Integrated Circuits,” (2017), arXiv:1708.02226.
395 T. C. White, J. Y. Mutus, I.-C. Hoi, R. Barends, B. Camp- 411 D. Rosenberg, D. Kim, R. Das, D. Yost, S. Gustavsson,

bell, Y. Chen, Z. Chen, B. Chiaro, A. Dunsworth, E. Jeffrey, D. Hover, P. Krantz, A. Melville, L. Racz, G. O. Samach, S. J.
J. Kelly, A. Megrant, C. Neill, P. J. J. O’Malley, P. Roushan, Weber, F. Yan, J. L. Yoder, A. J. Kerman, and W. D. Oliver,
D. Sank, A. Vainsencher, J. Wenner, S. Chaudhuri, J. Gao, “3D integrated superconducting qubits,” npj Quantum Informa-
and J. M. Martinis, “Traveling wave parametric amplifier with tion 3, 42 (2017).
josephson junctions using minimal resonator phase matching,” 412 N. T. Bronn, V. P. Adiga, S. B. Olivadese, X. Wu, J. M. Chow,

Applied Physics Letters 106, 242601 (2015). and D. P. Pappas, “High coherence plane breaking packaging
396 N. E. Frattini, V. V. Sivak, A. Lingenfelter, S. Shankar, and for superconducting qubits,” Quantum Science and Technology
M. H. Devoret, “Optimizing the nonlinearity and dissipation 3, 024007 (2018).
of a snail parametric amplifier for dynamic range,” Phys. Rev. 413 B. Foxen, J. Y. Mutus, E. Lucero, R. Graff, A. Megrant,

Applied 10, 054020 (2018). Y. Chen, C. Quintana, B. Burkett, J. Kelly, E. Jeffrey, Y. Yang,
397 E. Farhi, J. Goldstone, S. Gutmann, and M. Sipser, “Quantum A. Yu, K. Arya, R. Barends, Z. Chen, B. Chiaro, A. Dunsworth,
Computation by Adiabatic Evolution,” (2000), arXiv:0001106 A. Fowler, C. Gidney, M. Giustina, T. Huang, P. Klimov,
[quant-ph]. M. Neeley, C. Neill, P. Roushan, D. Sank, A. Vainsencher,
398 E. Farhi, J. Goldstone, S. Gutmann, J. Lapan, A. Lundgren, J. Wenner, T. C. White, and J. M. Martinis, “Qubit com-
and D. Preda, “A quantum adiabatic evolution algorithm ap- patible superconducting interconnects,” Quantum Science and
plied to random instances of an NP-complete problem.” Science Technology 3, 014005 (2018).
(New York, N.Y.) 292, 472–5 (2001). 414 T. G. McConkey, J. H. Béjanin, C. T. Earnest, C. R. H. McRae,
399 “The D-Wave 2000Q System,” https://www.dwavesys.com/d- Z. Pagel, J. R. Rinehart, and M. Mariantoni, “Mitigating leak-
wave-two-system. age errors due to cavity modes in a superconducting quantum
400 V. S. Denchev, S. Boixo, S. V. Isakov, N. Ding, R. Babbush, computer,” Quantum Science and Technology 3, 034004 (2018).
V. Smelyanskiy, J. Martinis, and H. Neven, “What is the Com- 415 “QCoDeS,” https://github.com/QCoDeS/Qcodes.

putational Value of Finite-Range Tunneling?” Physical Review 416 “PycQED,” https://github.com/DiCarloLab-

X 6, 031015 (2016). Delft/PycQED py3.
401 T. Albash and D. A. Lidar, “Demonstration of a scaling advan- 417 “qKIT,” https://github.com/qkitgroup/qkit.

tage for a quantum annealer over simulated annealing,” Phys. 418 “Labber Quantum,” http://quantum.labber.org/.

Rev. X 8, 031016 (2018). 419 Y. Liu, Quantum Feedback Control of Multiple Superconducting
402 J. King, S. Yarkoni, J. Raymond, I. Ozfidan, A. D. King, M. M. Qubits, Ph.D. thesis (2016).
Nevisi, J. P. Hilton, and C. C. McGeoch, “Quantum annealing 420 “Qiskit,” https://qiskit.org/.

amid local ruggedness and global frustration,” Journal of the 421 R. S. Smith, M. J. Curtis, and W. J. Zeng, “A Practical Quan-
 67

tum Instruction Set Architecture,” (2016), arXiv:1608.03355. 022316 (2005).

422 D. S. Steiger, T. Häner, and M. Troyer, “ProjectQ: An Open 429 C. J. Axline, L. D. Burkhart, W. Pfaff, M. Zhang, K. Chou,
Source Software Framework for Quantum Computing,” Quan- P. Campagne-Ibarcq, P. Reinhold, L. Frunzio, S. M. Girvin,
tum 2, 49 (2016), arXiv:1612.08091. L. Jiang, M. H. Devoret, and R. J. Schoelkopf, “On-demand
423 “Cirq,” https://github.com/quantumlib/Cirq. quantum state transfer and entanglement between remote mi-
424 J. R. McClean, I. D. Kivlichan, K. J. Sung, D. S. Steiger, Y. Cao, crowave cavity memories,” Nature Physics 14, 705–710 (2018).
C. Dai, E. S. Fried, C. Gidney, B. Gimby, P. Gokhale, T. Häner, 430 P. Kurpiers, P. Magnard, T. Walter, B. Royer, M. Pechal,

T. Hardikar, V. Havlı́ček, C. Huang, J. Izaac, Z. Jiang, X. Liu, J. Heinsoo, Y. Salathé, A. Akin, S. Storz, J.-C. Besse, S. Gas-
M. Neeley, T. O’Brien, I. Ozfidan, M. D. Radin, J. Romero, parinetti, A. Blais, and A. Wallraff, “Deterministic quantum
N. Rubin, N. P. D. Sawaya, K. Setia, S. Sim, M. Steudtner, state transfer and remote entanglement using microwave pho-
Q. Sun, W. Sun, F. Zhang, and R. Babbush, “OpenFermion: tons,” Nature 558, 264–267 (2018).
The Electronic Structure Package for Quantum Computers,” 431 J. Preskill, “Quantum computing and the entanglement fron-

(2017), arXiv:1710.07629. tier,” (2012), arXiv:1203.5813.

425 “The Microsoft Quantum Development kit,” 432 A. W. Harrow and A. Montanaro, “Quantum computational

https://github.com/Microsoft/Quantum. supremacy,” Nature 549, 203–209 (2017).

426 R. LaRose, “Overview and comparison of gate level quantum 433 A. M. Dalzell, A. W. Harrow, D. E. Koh, and R. L. La
software platforms,” Quantum 3, 130 (2019). Placa, “How many qubits are needed for quantum computa-
427 A. Kandala, K. Temme, A. D. Corcoles, A. Mezzacapo, J. M. tional supremacy?” (2018), arXiv:1805.05224.
Chow, and J. M. Gambetta, “Extending the computational 434 S. Boixo, S. V. Isakov, V. N. Smelyanskiy, R. Babbush, N. Ding,

reach of a noisy superconducting quantum processor,” (2018), Z. Jiang, M. J. Bremner, J. M. Martinis, and H. Neven, “Char-
arXiv:1805.04492. acterizing quantum supremacy in near-term devices,” Nature
428 S. Bravyi and A. Kitaev, “Universal quantum computation with Physics 14, 595–600 (2018).
ideal Clifford gates and noisy ancillas,” Physical Review A 71,