6 Exercise Set 6 - Gravitational Waves

6.1 GW space-time and (3) Killing vectors

Prompt: Show that the gravitational wave space-time (16.2 in Hartle) has three Killing
vectors: (0, 1, 0, 0), (0, 0, 1, 0), and (1, 0, 0, 1).
Solution: Memorize and write down the GW metric:

ds2 = −dt2 + [1 + f (t − z)]dx2 + [1 − f (t − z)]dy 2 + dz 2

ds is not a function of x or of y (the dx2 , and dy 2 terms do not count). Therefore, the x
and y vectors
ξxµ = (0, 1, 0, 0)
and
ξyµ = (0, 0, 1, 0)
are Killing vectors.

And for the third (1, 0, 0, 1) vector: Make the change of coordinate

t→t+δ

z →z+δ
The deltas will appear in the f∗ (t − z) terms, but they will cancel out. Thus the 3rd
killing vector is
ξ3µ = (1, 0, 0, 1)
... ... ... ′µ
the metric does not depend on z. So, in the coordinate system (u, x, y, z), we have the Killing vector ξ = (0, 0, 0, 1).
µ
(HOW??? ??? ???) We transform this vector to the original coordinate system, and we get ξ = (1, 0, 0, 1).

Alternate solution: (1, 0, 0, 1) is a Killing vector because the metric is invariant under the transformation (t, x, y, x) → (t + δ, x, y, z + δ).

This is. xµ → xµ + δξµ , where ξµ is the Killing vector.

6.2 GW on one test mass

Prompt: Consider a test mass at rest at position x = (X, Y, Z) in flat spacetime. A
gravitational wave (16.2 in Hartle) passes over the test mass. Show that to leading order
in the amplitude of the wave the coordinate position of the test mass remains (X, Y, Z).
Compare the solution below with the section 16.2 of Hartle’s Gravity.
Solution:
Consider the test mass at (X, Y, Z). If it moves, let’s call its new coordinates (x + δx, y +
δy, z + δz) = (x′ , y ′ , z ′ ). ”the amplitude of the wave the coordinate position” means
movement, which means geodesic equation:

d 2 xi α
i dx dx
β
= −Γαβ
dτ 2 dτ dτ
We’re considering small changes in coordinates, thus:

d2 δxi α
i dδx dδx
β
= −δΓαβ
dτ 2 dτ dτ
38
The δs in the right-hand terms are approximately negligible,

d2 δxi α

x dδ
x
i dδ
β
= −δΓαβ
dτ 2 dτ dτ
d2 δxi i dx dx
α β
= −δΓ αβ
dτ 2 dτ dτ
2 i
d δx
= −δΓiαβ uα uβ
dτ 2
At rest, we have the general vector uαrest = (1, 0, 0, 0)

d2 δxi
= −δΓitt (1)
dτ 2
Also, given a gravitational wave, we have

d2 δxi i i 1 iα
= −δΓ tt = dΓαβ = η (∂α hµα + ∂β hµβ − ∂i hαβ )
dτ 2 2
I think there are many 0 terms as we go from the above general equation to the below
particular equation,
1
δΓitt = η iµ (2 ∂t htµ − ∂µ htt + 0 + 0...)
2
(WHY DOES???) This equals zero,
1
δΓitt = η iµ (2 ∂t htµ − ∂µ htt ) = 0
2
So,
d2 δxi
=0
dτ 2
Take an integral,
dδxi
=C
dτ
(WHY IS???) This being zero,
dδxi
=0
dτ
Also implies constant, unchanged coordinates, so

X ′ , Y ′ , Z ′ = X, Y, Z

6.3 GW on two test masses

Prompt: Consider the gravitational wave in (16.2 in Hartle) and two test masses, one
at the origin and the other at a location (X, Y, Z) in the Cartesian coordinates used in
(16.2). Show that the 1change in distance between the masses produced by the wave is
given by
1 L∗
Z
δL(t) = L − L∗ = hij (t − λnz )ni nj dλ
2 0

39
Where  
i X Y X
n = , ,
L∗ L∗ L∗
Here ni is the unit tangent vector to the straight-line path between the test masses and
L∗ is the unperturbed distance between them
Solution: Memorize and begin with parametrizing the straight line joining the origin
with the point (X, Y, Z) as

xµ (λ) = (0, λnx , λny , λnz ) → ẋµ = (0, nx , ny , nz )

from
⃗x = (t, x, y, z)
The unperturbed distance L∗ is
√
L∗ = X2 + Y 2 + Z2

The parameter λ is the space inside of the 0 → L (we’ll integrate it calculate it) and it
takes the values 0 ≤ λ ≤ L∗ .
Memorize and use the normal/generic GW Minkowski perturbation metric:

gµν = ηµν + hµν (t − z)

Note: the hµν is a function of (t − z), it is NOT multiplied by the factor (t − z) Memorize
and use the definition of L (geodesic equation):
Z L∗
p
L= gµν ẋµ ẋν dλ
0
Z L∗ q
L= (ηµν + hµν (t − z))ẋµ ẋν dλ
0
Convert index notation, so before µν → ij now. Use

ẋi = ni

ηij ẋi ẋj = 1

||ni || = 1
The movement of the wave is entirely in the z direction, so the general z and ηµν term
take the form
z = λnz
ηµν = 1
Z L∗ q
L= 1 + hij (t − λnz )ni nj dλ
0
Z L∗ p
L= 1 + hij ni nj dλ
0

Within the square root, |hij (t − λn )ni nj | << 1. Use a Taylor expansion for the square
z

root; which is in general

√ x
1+x∼1+
2
40
and in this particular instance
1
q
1 + hij (t − λnz )ni nj ∼ 1 + (hij (t − λnz )ni nj )
2
and Z L∗ Z L∗
1
q
L= z i j
1 + hij (t − λn )n n dλ ∼ [1 + (hij (t − λnz )ni nj )]dλ
0 0 2
Z L∗
1
L= [1 + (hij (t − λnz )ni nj )]dλ
0 2
Z L∗ Z L∗
1
L= (1)dλ + (hij (t − λnz )ni nj )dλ
0 0 2
Z L∗
1
L = (L∗ − 0) + (hij (t − λnz )ni nj )dλ
0 2
Z L∗
1
L − L∗ = (hij (t − λnz )ni nj )dλ
2 0
1 L∗
Z
L − L∗ = hij (t − λnz )ni nj dλ
2 0
1 L∗
Z
L − L∗ = hij ni nj dλ
2 0
Shown. Now continue this by solving for a reasonable hij (first half of next problem) and
substituting it into this equation (second half of next problem).

6.4 GW Polarization
δL(t)
Prompt: Express for a ”+” polarized wave of definite frequency ω and amplitude
L∗
a.
Solution: Memorize beginning with hij
 
htt htx hty htz
hxt hxx hxy hxz 
hij = 
hyt hyx hyy hyz 


hzt hzx hzy hzz

Let the z-axis be linearized and aligned with the direction of the wave’s propagation. This
means    
htt htx hty htz 0 0 0 0
hxt hxx hxy hxz  0 hxx hxy 0
hij = 
hyt hyx hyy hyz  = 0 hyx hyy 0
  

hzt hzx hzy hzz 0 0 0 0

   
0 0 0 0 0 0 0 0
0 hxx hxy 0 0 f+ (t − z) fx (t − z) 0
hij = 0 hyx hyy 0 = 0 fx (t − z) −f+ (t − z) 0
  

0 0 0 0 0 0 0 0

41
Use the ”+” wave polarization
   
0 0 0 0 0 0 0 0
0 f+ (t − z) fx (t − z) 0 0 f+ (t − z)
  0 0
hij = 
0 fx (t − z) −f+ (t − z) = 
0  0 0 −f+ (t − z) 0
0 0 0 0 0 0 0 0

Use the ω and a terms

   
0 0 0 0 0 0 0 0
0 f+ (t − z) 0 0 0 a cos(ω(t − z))
  0 0
hij =  = 
0 0 −f+ (t − z) 0 0 0 −a cos(ω(t − z)) 0
0 0 0 0 0 0 0 0

or more compactly,
hxx = −hyy = a cos(ω(t − z))
Turn (t − z) into (t − λnz ) by aligning z-axis to wave direction. Combine this and the
above with this solution from the last exercise
1 L∗ 1 L∗
Z Z
z i j
L−L∗ = hij (t−λn )n n dλ = [hxx nx nx +hxy nx ny +hyy ny ny +hxz nx nz +...]dλ
2 0 2 0
to get:
1 L∗
Z
L − L∗ = (a cos(ω(t − z)))(t − λnz )ni nj dλ
2 0
1 L∗
Z
δL = (a cos(ω(t − z)))(t − λnz )ni nj dλ
2 0
simplify; only two of the n’s are non-zero (nxx and nyy )
Z L∗
1
δL = (a cos(ω(t − λnz )))((nx )2 − (ny )2 ) dλ
2 0
Z L∗
1
δL = ((nx )2 − (ny )2 ) (a cos(ω(t − λnz ))) dλ
2 0
Z L∗
a
δL = ((nx )2 − (ny )2 ) (cos(ω(t − λnz ))) dλ
2 0
let,
du
u = ω(t − λnz ), so du = −ωnz dλ and dλ = −
ωnz
substitute, Z L∗
a du
δL = ((nx )2 − (ny )2 ) (cos(u))
2 0 −ωnz
Z L∗
a
δL = − ((nx )2 − (ny )2 ) (cos(u)) du
2ωnz 0
L∗
a x 2 y 2

δL = − z
((n ) − (n ) )(sin(u))
2ωn 0

42