GR HW 1

Due Jan 30 at 2pm, in class

Goals: Lorenz transformations, Four-vector and four-tensor notation, Relativistic kinematics,

Rest-frame, Energy-momentum conservation, Levi-Civita tensor.

1. Two events xµA and xµB are space-like separated. Find the boost to a Lorentz frame in which
the two events are simultaneous.
2. Spaceship is coming back to Earth at 0.995c. During a TV contact, people on Earth notice
that time runs faster onboard (by how much?). How come the astronauts become younger
than people on Earth?
3. An ideal photon spaceship (transforming fuel mass to energy of ideally collimated photons
with 100% efficiency) goes to the center of the galaxy (10 kpc from Earth) and back. It starts
from rest and finishes at rest. The accelerations and decelerations are equal to a = 10m/s2 .
Calculate:
a. The trip time by Earth and onboard clocks.
b. The initial mass of the fuel M . (The mass of the ship without fuel plus the mass of the
astronauts is m.)
4. The neutral Sigma baryon, Σ0 , with mass mΣ , decays into a Lambda baryon, Λ, with mass
mΛ , and a massless photon.
(a) Find the energy of the photon in the frame in which the Σ0 is at rest.
(b) Find the energy of the photon in the frame in which the Λ is at rest.
Hint: it simplifies the algebra to use four-vectors.
5. φ, Aµ , T µν are scalar, vector and tensor. Which of the following equations are covariant
a. φ = A0
b. φ = Aµ Aµ
c. φ = A0 A0
d. φ = Tµν T µν
e. Tµν = T νµ
f. Tµν = Tνµ
g. T µν = Aµ + Aν
h. Tµν = −Tνµ
i. Tνµ = −Tµν
j. T µν = Aµ Aν
k. φ = det T µν
l. φ = det Tνµ
6. In d dimensions
εµ1 ···µd εν1 ···νd Aνµ11 · · · Aνµdd = c det Aνµ . (1)
Find c.

1
 GR HW 2
Due Feb 6 at 2pm, in class

Goals: Symmetries and conserved quatities, Stress-energy tensor, Representations of Poincaré,

Forces mediated by particles

1. For a point-like particle derive conserved quantities corresponding by Noether theorem to

invariance under Lorentz transformations X µ → X µ + ωνµ X ν . Do it following the procedure
we outlined in the class, namely by considering Lorentz transformations with time dependent
parameters ωνµ → ωνµ (σ).
2. Derive the equations of motion following from the action
dX µ
Z
1  
SPolyakov = dσe−1 Ẋ µ Ẋµ − e2 m2 , Ẋ µ = ,
2 dσ
where e (which is called the worldline metric) is an additional dynamical variable. Are
these equations equivalent to the equation of motion for a point-particle? Note that in this
action (unlike the Nambu-Goto action), one can set m = 0. Does it reproduce what you
expect for a massless particle? What is the transformation of e under reparametrizations of
σ?
3. In addition to the physical time translation invariance X 0 → X 0 +  the action for a point-
like particle is obviously invariant also under the worldline time translation (a special case of
worldline reparametrization),
X µ (σ) → X µ (σ + ) .
Normally, by Noether theorem one expects to obtain the conserved “worldline energy”, cor-
responding to this symmetry. Calculate this energy.
4. The stress-energy tensor of a relativistic perfect fluid is given by
Tνµ = (ρ + p)uµ uν + pδνµ (1)
where ρ, p and uµ are respectively fluid density, pressure and 4-velocity. uµ
is a unit time-like
µ
vector uµ u = −1. For a non-relativistic fluid ρ = mn + ρK where m is the mass of fluid
particles and n is their number density and ρK is the density of kinetic plus potential energy.
In this limit p ∼ ρK ∼ mnv 2  mn where v is the fluid 3-velocity. The equation of state
p(n) fixes the pressure in terms of number density. Take the non-relativistic limit of fluid
stress-energy conservation ∂µ Tνµ = 0, to find the time-evolution of ρ and v.
5. Find the number of degrees of freedom (polarizations) of dilaton (massless scalar), photon
(massless spin 1), and graviton (massless spin 2) in d = 5 spacetime dimensions. (∗ Are there
photons or gravitons in d = 2 or d = 3?)
6. Calculate the differential scattering cross-section in attractive 1/r potential with strength C.
Use Born approximation.
7. Calculate the differential scattering cross-section for two non-relativistic particles of mass
m, M (m  M ), due to the tree-level exchange √ of aRscalar field ϕc coupled universally to the
trace of the stress-energy tensor, i.e. Sint = 4πG d xϕc Tµµ . Is there a choice of C in the
4

previous problem that gives the same answer as what you find here in the rest frame of the
mass-M particle?

1
 GR HW 3
Due Feb 13 at 2pm, in class

Goals: Gravitational redshift, Scalar gravity waves, Small angle scattering

1. Twin brothers are separated at birth: one stays at sea level the other lives at a space station,
freely orbiting at radius 20,000km. Which one will outlive his brother (will see him die) and
by how much? (Earth mass 5.97 × 1027 gr.)

2. What’s the temperature difference (in thermodynamic equilibrium at room temperature)

between the floor and the ceiling (3m high room, due to gravity)?

3. ∗ Is there any dilaton radiation if a photon of energy E is trapped inside a box of mass M  E
and size R? How much does the
R box weigh on average? (Hint: Use stress-energy conservation
to determine the time-average d3 xTµν t for (µ, ν) = (0, 0) as well as (µ, ν) = (i, j) in a
compact periodic system.)
If there is radiation calculate the spectrum d hLi /dω for ω  1/τ and estimate the total
luminosity. τ  R is the microscopic time it takes for the photon to bounce.

4. A body of mass m  M passes by a body of mass M . In the small angle scattering approxi-
mation, estimate total radiated energy. Assume non-relativistic velocity.

5. Estimate how long it would take the Earth to fall down on the Sun due to emission of
gravitational waves (in scalar gravity).

1
 GR HW 4
Due Feb 20 at 2pm, in class

Goals: Field of static sources, Emission of GWs, Motion in gravitational field, Light bending

1. a)∗ Show that at a large distance r from a massive body of mass M , the hµν field is given by
rg rg
h00 = , hij = δij (1)
r r
where rg = 2GM/c2 .
b) Calculate the small angle scattering (scattering angle χ in lowest order in 1/b, b is the
impact parameter) for a non-relativistic and an ultrarelativistic particles.
In particular, calculate light deviation by the Sun, for a light ray just grazing the surface
of the Sun (solar mass M = 1.99 × 1033 gr, radius b = 700, 000 km, give your answer in
arcminutes).
c) For a non-relativistic particle of mass m  M , estimate (∗ or calculate if you feel energetic)
total radiated energy.

2. Do you emit gravitational waves when you move your arm? (Hint: this problem does require
a certain numerical estimate before you can tell.)

3. ∗ Derive Eqs.(1-4) of http://arxiv.org/abs/1207.7176. Calculate the amplitude h0 for

a neutron star with mass 2 solar-mass, eccentricity 0.1, radius 10km, spinning at 100Hz,
distance to the Earth 1kpc, and 0 inclination angle.

4. Consider a burst of GWs of the form hyy = −hzz = A cos (ω(t − x)) shining on LIGO mirrors,
perfectly free masses in y direction, initially separated by ∆y = L  c/ω. Calculate the time
it takes, measured by the clock attached to one of the mirrors, for the laser light to go to the
other mirror and come back (A = 10−20 , ω = 100Hz, L = 4km).

5. In the above problem, what is the distance between the two masses as measured by good
rulers (rulers made of microscopic elements that are strongly bound together)? If the two
masses are attached by a soft spring with 2k/m ≈ ω 2 does it resonate?
FYI: Joseph Weber proposed a method for detecting GWs based on this idea.

1
 5

I. INTRODUCTION dramatically with the time baseline, and the computa-

tional requirements of the search increase rapidly. Even
A promising class of sources for detectable gravita- with a coherent time baseline of just a few days, a
tional wave signals are rapidly rotating neutron stars wide-frequency-band all-sky search is computationally
with non-axisymmetric deformations [1–5]. Such objects extremely challenging. It becomes completely unfeasi-
are expected to emit long-lived continuous-wave (CW) ble if one considers instead time baselines on the order
signals. In the rest frame of the neutron star, these waves of months.
have a constant amplitude and are quasi-monochromatic As is often the case with computationally bound prob-
with a slowly decreasing intrinsic frequency. They are lems, hierarchical approaches have been proposed [18–
received at Earth-based detectors with a Doppler modu- 20]. In these strategies, the entire data set is split into
lation due to the relative motion between the source and shorter segments. Each segment is analyzed coherently,
the detector. Consequently the observed phase evolution and afterwards the information from the different seg-
depends on the intrinsic signal frequency, the first fre- ments is combined incoherently (which means that the
quency time-derivative (also called spindown), and the phase information is lost). The amplitude sensitivity
source sky position; these parameters shall collectively grows at best with the fourth root of the number of
be called the phase evolution parameters. While using segments. Such methods have been used in previous
higher order frequency derivatives could potentially im- wide-parameter-space searches published by the LIGO
prove the astrophysical detection efficiency in a part of and Virgo collaborations [21–26].
the parameter space (see Sec. III), we shall not consider A subset of these searches [21, 22, 25, 26] used seg-
them in this paper for computational reasons. Finally, ments sufficiently short (1 800 s) that the signal remains
the received signal has a time-dependent amplitude mod- within a single Fourier frequency bin in each segment.
ulation due to the (time-dependent) relative geometry of In these cases, a simple Fourier transform suffices for
the wave and the detector. each segment. Three different variants of such methods
The previous two decades have seen the construction have been developed that combine the results from the
and operation of several kilometer-scale laser interfero- different short segments incoherently: the “stack-slide”,
metric gravitational wave detectors [6–11]. The detectors the “Hough-transform” and the “PowerFlux” schemes.
and the data analysis tools have steadily improved over The stack-slide procedure [18, 20] averages the normal-
this period. These have made it possible to search for ized power from the Fourier transform of 30-minute seg-
various gravitational wave signals with ever-improving ments (Short time baseline Fourier Transform, SFT for
sensitivity. In this paper we focus on data from the fifth short) of the calibrated detector strain data. The Pow-
science run (S5) of the LIGO (Laser Interferometer Grav- erFlux scheme [22, 25] can be seen as a variation of the
itational wave Observatory) detectors, collected between stack-slide method, where the power is weighted before
the GPS times of 815 155 213 s (Fri Nov 04 16:00:00 UTC summing. The weights are chosen according to the detec-
2005) and 875 145 614 s (Sun Sep 30 00:00:00 UTC 2007). tor noise and antenna pattern to maximize the signal-to-
The LIGO network [6] consists of three detectors: one at noise ratio (SNR). The Hough-transform method [19, 27]
Livingston, Louisiana, USA, with an arm length of 4 km sums weighted binary counts, depending upon whether
(L) and two in the same vacuum envelope at Hanford, the normalized power in an SFT bin exceeds a certain
Washington, USA, with arm lengths of 4 km (H) and threshold.
2 km, respectively. Only data from H and L detectors As the segment duration is increased, it becomes nec-
are used in this paper. The Virgo and GEO 600 detec- essary to account for signal modulations within each seg-
tors also collected data during the same time interval but ment by computing the F-statistic over a grid in the
are not used in this analysis, which is optimized for two space of phase evolution parameters. This results in a sig-
detectors with similar sensitivities. nificant increase in the computational requirements of the
A coherent strategy for extracting faint CW signals search. The distributed volunteer computing project Ein-
buried in noisy data using standard maximum-likelihood stein@Home [28] has been created to address this need.
techniques in the presence of “nuisance parameters” was Two previous papers [23, 24] report on results of such
derived in [12]. The resulting detection statistic is the CW searches from the fourth LIGO science run and from
so-called F-statistic, which has since been generalized to the first two months of S5, respectively. The method used
the case of multiple detectors [13, 14]. The F-statistic was based on the computation of the coherent F-statistic
has also been shown to arise as a special case in a more on data segments from either the H or L detectors sepa-
general Bayesian framework [15]. Using the F-statistic rately, and only parameter space points with values of 2F
means that we need to search explicitly only over the larger than 25 were returned back to the Einstein@Home
phase evolution parameters. server for further inspection. The threshold value of 25
Coherent wide-parameter-space searches utilizing the limited the ultimate sensitivity of that search: if a sig-
F-statistic have been carried out since the second LIGO nal was not loud enough to surpass that threshold on at
science run [16, 17]. The amplitude sensitivity of this least some of the segments it would not be detected. The
type of search improves as the square root of the time threshold value was set by bandwidth constraints on the
baseline. However, the template bank spacing decreases size of the results file uploaded back to the server by the
 6

host, i.e. on the maximum number of significant points other axes. The amplitude is given by
that could be returned. These results were subsequently
4π 2 G Izz f 2 ε
combined by a coincidence scheme, performed offline in h0 = , (4)
the post-processing phase. In contrast, in the search pre- c4 d
sented here, the combination of the results from the co- where f is the frequency of the emitted GW signal (which
herent searches takes place directly on the host machines is also twice the rotational frequency of the star), G is
using a Hough-transform scheme. This makes it possible Newton’s constant, c is the speed of light and d is the
to use a much lower threshold on 2F, equal to 5.2, that distance to the star. The distribution of ε for neutron
defines the parameter space points to be passed on to the stars is uncertain and model dependent since the break-
Hough-transform. Moreover, in this search, data from ing strain for a neutron star crust is highly uncertain (see
the H and L detectors are coherently combined [13, 14]. e.g. [2, 29–31] for further discussion).
Finally, more data was searched in this analysis com- Energy loss from the emission of gravitational and/or
pared to any previous Einstein@Home search. The Ein- electromagnetic waves, as well as possible local acceler-
stein@Home runs presented here refer to searches based ation of the source, causes the signal frequency arriving
on the first (S5R3) and second year of S5 LIGO data. at the solar system to evolve. To first order, it can be
This latter search was run on Einstein@Home in two sep- expressed as
arate steps, called S5R5 and S5R6. Since the S5R6 run
fˆ(τ ) = f0 + f˙(τ − τ0 ), (5)
used the same data as S5R5, but extended the search re-
gion above 1 kHz, in this paper we simply refer to these where τ is the arrival time of a wavefront at the solar sys-
two runs as S5R5. tem barycenter (SSB), f0 is the frequency at a fiducial
The paper is structured as follows. In Sec. II, we reference time τ0 , and f˙ denotes the first time derivative
briefly review the Hough-transform method. Section III of the frequency. The astrophysical implications of ig-
describes the Einstein@Home distributed search used to noring higher order derivatives in this Taylor expansion
analyze the data set. Section IV gives a detailed descrip- will be discussed later. The phase of the signal, φ(t), fol-
tion of the S5R5 post-processing, which is based on the lows directly from the frequency evolution with an initial
pioneer S5R3 post-processing (described in Appendix B). phase φ0 at the reference time.
Upper limit computations from the more-sensitive S5R5 As the detector on the Earth moves relative to the SSB,
data are provided in Sec. V. The study of some hardware- the arrival time of a wavefront at the detector, t, differs
injected signals is presented in Sec. VI. In Sec. VII we from the SSB time τ :
make some concluding remarks. ~r(t) · ~n
τ (t) = t + + ∆E − ∆S . (6)
c
II. THE DATA ANALYSIS METHOD Here ~r(t) is the position vector of the detector in the
SSB frame, and ~n is the unit vector pointing to the neu-
A. The waveform model tron star; ∆E and ∆S are respectively the relativis-
tic Einstein and Shapiro time delays [32]. In standard
equatorial coordinates with right ascension α and decli-
Let us begin by briefly describing the standard signal
nation δ, the components of the unit vector ~n are given
model for CW signals. In the rest frame of the neutron
by (cos α cos δ, sin α cos δ, sin δ).
star, the gravitational wave signal is elliptically polarized
Ignoring the relativistic corrections, the instantaneous
with constant amplitudes A+,× for the two polarizations
frequency f (t) of a CW signal, as observed at time t
h+,× (t). Thus, we can find a frame such that
by a detector on Earth, is described by the well-known
h+ (t) = A+ cos φ(t) , h× (t) = A× sin φ(t) . (1) Doppler shift equation:
~v (τ ) · ~n
The two amplitudes are related to an overall amplitude f (t) = fˆ(τ ) + fˆ(τ ) , (7)
c
h0 and the inclination angle ι between the line of sight
to the neutron star and its rotation axis where ~v (τ ) is the detector velocity with respect to the
SSB frame; ~v (τ ) is the sum of two components, from the
1 yearly Earth motion around the Sun (~vy ) and from the
A+ = h0 (1 + cos2 ι) , A× = h0 cos ι . (2)
2 rotation of Earth around its axis (~vd ).
Finally, the received signal at the detector is
The value of h0 is model dependent. For emission due to
non-axisymmetric distortions, the amplitude h0 depends h(t) = F+ (t; ~n, ψ)h+ (t) + F× (t; ~n, ψ)h× (t), (8)
on the ellipticity ε of the star defined as
where F+,× are the detector beam pattern functions
|Ixx − Iyy | which depend on the sky position ~n and the relative po-
ε= . (3) larization angle ψ of the wave-frame [12, 33]. There are
Izz
thus altogether eight signal parameters, which include
Here Izz is the principal moment of inertia of the star, the four phase evolution parameters (f0 , f˙, α, δ), and four
and Ixx and Iyy are the moments of inertia about the other parameters (h0 , ι, ψ, φ0 ).
 GR HW 5
Due Feb 27 at 2pm, in class

Goals: Covariant stress-energy tensor, Conserved quantities on curved spacetime, Non-Cartesian

coordinates

µ
1. We know that Tν;µ = 0 does not express any conservation law for a generic spacetime.
But in a constant gravitational field, there does exist a conserved energy E. Express E
in terms of Tνµ , and prove its conservation in a constant gravitational field. (Note: for
spacetimes invariant under spatial translations momentum is conserved, rotations give angular
momentum conservation.)

2. We know that energy is conserved along the geodesic if the metric tensor does not depend on
time. If the metric does not depend on x1 , is there a corresponding conservation law?

3. Find the energy momentum tensor of electromagnetic field by varying the action
Z
1
S=− d4 xF 2 , Fµν = Aµ;ν − Aν;µ (1)
4
with respect to the metric.

4. When calculating the energy momentum tensor of electromagnetic field by varying with re-
spect to the metric, what should you keep fixed Aµ or Aµ ?

5. ∗ Calculate the total energy (gravitational field plus matter) for a Schwarzschild black hole.
Schwarzschild metric is
dr2
ds2 = −(1 − rg /r)dt2 + + r2 (dθ2 + sin2 θdφ2 ). (2)
(1 − rg /r)

6. For two-dimensional flat space, ds2 = dr2 + r2 dθ2 , calculate Christoffel symbols Γijk for all
i, j, k = r, θ. Using the obtained Christoffel symbols, calculate the divergence of a vector field
∇i Ai . Explain why your expression differs from the standard textbook expression
∇ · A = 1r ∂r (rAr ) + 1r ∂θ Aθ

7. Write the (2 + 1)d Minkowski metric in radial coordinates t, r, θ. Derive the non-relativistic
version of geodesic equation with a force-field f µ . What is the relation between f i and
components of the Newtonian force F in radial coordinates?

8. ∗ Find the light cone (null hypersurface) with a vertex at t = x = y = z = 0 (the light cone
is formed by all light rays passing through the vertex) in the gravitational field

hxx = x2 /a2 , hyy = y 2 /b2 , hzz = z 2 /c2 , (3)

where a, b, c are some constants.

1
 GR HW 6
Due Mar 6 at 2pm, in class

Goals: Coordinate systems, Curvature, Geodesic deviation

1. Calculate components of d-dimensional Euclidean space metric in spherical coordinates r, θ1 , · · · , θd−2 , φ.

√
Calculate the corresponding integration measure gdd x. (Hint: first do it for d = 3.)

2. Consider a two-dimensional sphere of radius R embedded in the Euclidean space,

x2 + y 2 + z 2 = R 2 . (1)

As we discussed, one can construct a coordinate patch (x, y) covering all of the sphere, but
the north pole, by considering a stereographic projection on the plane

z = −R. (2)

Calculate components of the induced metric on the sphere in stereographic coordinates.

3. Trieste latitude is 45◦ . A vector is applied in Trieste, pointing towards the north pole, and
then is parallel transported along the 45◦ latitude, moving west, back to Trieste. Which way
will the vector turn, and by what angle?
√
4. Calculate the extrinsic curvature (the proper acceleration aµ aµ ) of lines of constant φ and
lines of constant θ in S 2 , and of lines of constant r and lines of constant φ in R2 .

5. Lobachevski plane is defined by the metric

dx2 + dy 2
ds2 = , x > 0, (3)
x2
a) Find the geodesics by direct minimization of the length. Draw the geodesics on the xy
plane.
b) Show that Rµναβ ∝ x−4 , Rµν ∝ x−2 and R = const. Using this, find all these quantities by
doing parallel transport along a convenient contour.

6. Solve problem 1.(a) of section 14 of the notes.