L∗

a x 2 y 2 z

δL = − ((n ) − (n ) )(sin(ω(t − λn )))
2ωnz 
0
a
δL = − ((nx )2 − (ny )2 )(sin(ω(t − (L∗ )nz )) − sin(ω(t − (0)nz )))
2ωnz
a
δL = − ((nx )2 − (ny )2 )(sin(ω(t − (L∗ )nz )) − sin(ωt))
2ωnz
a
δL = ((nx )2 − (ny )2 )(sin(ωt) − sin(ω(t − (L∗ )nz )))
2ωnz
a
δL = z
[(nx )2 − (ny )2 ][sin(ωt) − sin(ω(t − L∗ nz ))]
2ωn
z
Let Z = L∗ n
a
δL = z
[(nx )2 − (ny )2 ][sin(ωt) − sin(ω(t − Z))]
2ωn
ni X Y Z
Use xi = (X, Y, Z) and ni = L∗
, which is: nx = L∗
, ny = L∗
, and nz = L∗

a X Y
δL = Z
[( )2 − ( )2 ][sin(ωt) − sin(ω(t − Z))]
2ω L∗ L∗ L∗

Now make the ratio,

a
δL(t) 2ω LZ
[( LX∗ )2 − ( LY∗ )2 ][sin(ωt) − sin(ω(t − Z))]
∗
=
L∗ L∗
δL(t) 1 a X Y
= ( )[ Z [( )2 − ( )2 ][sin(ωt) − sin(ω(t − Z))]]
L∗ L∗ 2ω L∗ L∗ L∗
δL(t) 1 aL∗ X 2 Y
= ( )[ [( ) − ( )2 ][sin(ωt) − sin(ω(t − Z))]]
L∗ L∗ 2ωZ L∗ L∗
δL(t) a X Y
= [( )2 − ( )2 ][sin(ωt) − sin(ω(t − Z))]
L∗ 2ωZ L∗ L∗
δL(t) a 1
= [ (X 2 − Y 2 )][sin(ωt) − sin(ω(t − Z))]
L∗ 2ωZ L2∗
δL(t) a(X 2 − Y 2 ) 1
= [ ][sin(ωt) − sin(ω(t − Z))]
L∗ 2ωZL2∗ 1
δL(t) a(X 2 − Y 2 )
= [sin(ωt) − sin(ω(t − Z))]
L∗ 2ωZL2∗
Use this equation with a specified initial condition (ring of resting particles) to solve
exercise 5

43
6.5 GW deformation (circle to ellipse)
Prompt: The equation for an ellipse is x2 /â2 + y 2 /b̂2 = 1 where â is the semi-major
axis and b̂ is the semi-minor axis if â > b̂. Show that an initial circle of test particles
distorts into an ellipse according to (16.13 in Hartle) to lowest order in a and compute
the semi-major and semi-minor axes as a function of time
Solution Consider particles initially in the circle at

x2 + y 2 = R 2 , z=0

and a gravitational wave with (for instance) “+” polarization in the z direction. According
to the previous exercise, the distance between those particles and the origin (0, 0, 0) when
the gravitational wave passes is written from the above exercise
δL(t) δL(t) a(X 2 − Y 2 )
= = [sin(ωt) − sin(ω(t − Z))]
L∗ R 2ωZL2∗
memorize the simplification Z → 0 ... ... ...
δL(t) δL(t) a(x2 − y 2 )
= = cos(ωt)
L∗ R 2R2
Where does this (below) equation belong???

δR a(X 2 − X 2 )
= cos(ωt)
R 2R2
Now begin the derivation knowing that we’re going from a circle x2 + y 2 = R2 to an ellipse
of the shape
x2 + y 2 = (R + δR)2
This implies a change in the length of size δL, or

L∗ → L∗ + δL

Next, use the equation of the above equation to find

a(X 2 − Y 2 )

δR 
= [sin(ωt) − sin(ω(t − Z))]
R f ixed z>0 2ωZL2∗

a(X 2 − Y 2 )

δR 
= [sin(ωt) − sin(ω(t − Z))]
R f ixed z>0 2ωZL2∗
Now let everything go to infinitesimally small quantities
a(X 2 − Y 2 )
   
δR
lim = lim [sin(ωt) − sin(ω(t − Z))]
Z→0 R Z→0 2ωZL2∗

a(X 2 − Y 2 )
   
δR sin(ωt) − sin(ω(t − Z))
lim = lim
Z→0 R 2ωL2∗ Z→0 Z
The far right term is the definition of a derivative
a(X 2 − Y 2 ) d
 
δR 
= (sin(ωt))
R Z=0 2ωL2∗ dZ

44
 a(X 2 − Y 2 )
 
δR 
 = cos(ωt)
R 
Z=0 2ωL2∗
Now continue with... ... ...
a(x2 − y 2 )
x2 + y 2 = (R + δL(t))2 = R2 [1 + ( cos(ωt)) ]2
2R2
1 2 2 a(x2 − y 2 )
2
(x + y ) = [ 1 + ( 2
cos(ωt)) ]2
R 2R
Maybe at this point, expand the square, and drop the 2nd order (3rd) term, then take
the square root ASK Ludovico Machet!
r
1 2 2) = [ 1 + (
a(x2 − y 2 )
(x + y cos(ωt)) ]2
R2 2R2
Do Taylor series expansion
r  2 2
2
1 2 a(x − y )
2
(x + y 2 ) ≈ 1 + 2 2
cos(ωt) + (...2 )2 + ...
R 2R

Assume higher order terms are negligible

r  2 2
2
1 2 a(x − y )
(x + y 2 ) ≈ 1 + 2 cos(ωt)
R2 2R2
r
2 2
  
1 2 1 a(x − y )
(x + y 2 ) ≈ 1 + 2 cos(ωt)
R2 2 2R2
2 2
 
1p 2 a(x − y )
(x + y 2 ) = 1 + cos(ωt)
R 2R2
2 2
 
1p 2 a(x − y )
(x + y 2 ) = 1 + cos(ωt)
R 2R2
2 2
  
2 1
p a(x − y )
R (x2 + y 2 ) = 1 + cos(ωt)
R 2R2
a(x2 − y 2 )
 
p 2
2 2
R (x + y ) = R + cos(ωt)
2
This still doesn’t go towards the answer...
working backwards from the answer:

x2 + y 2 = R2 + a cos(ωt)(x2 − y 2 )

x2 + y 2 = R2 + cos(ωt)(x2 a − y 2 a)
x2 + y 2 = R2 + x2 a cos(ωt) − y 2 a cos(ωt)
x2 − x2 a cos(ωt) + y 2 + y 2 a cos(ωt) = R2
(x2 − x2 a cos(ωt)) + (y 2 + y 2 a cos(ωt)) = R2
x2 (1 − a cos(ωt)) + y 2 (1 + a cos(ωt)) = R2

45
Do we assume that x and y are small and that the x4 and y 4 terms will be negligible???
No... This would lead to a perfect circle all the time.
Somehow the above becomes
1
[ x2 (1 − a cos(ωt)) + y 2 (1 + a cos(ωt)) ] = 1
R2
This is of the ellipse form, x2 /a2 + y 2 /b2 = 1 which means that

a2 = R2 /(1 − a cos(ωt))

b2 = R2 /(1 + a cos(ωt))
and p
a = R/ (1 − a cos(ωt))
p
b = R/ (1 + a cos(ωt))
For is the 1 term dominates, use Taylor expansion and get
1
a ∼ R/(1 − a cos(ωt))
2
1
b ∼ R/(1 + a cos(ωt))
2
A similar logic applies to the other ”x” polarization.

6.6 Rotation as superposition for GWs

Prompt: In Section 16.3 in Hartle we produced a gravitational wave with × polarization
by rotating the + polarization (16.2) by 45o . Show that a rotation by an arbitrary angle
theta doesn’t give another independent solution but rather one that could be written as
a superposition of + and ×. This is one way of seeing that there are only two linearly
independent polarization of a gravitational wave.
Solution. Start with the memorized rotation matrix in the z direction (in the end,
nothing (important) changes
... ... ...

6.7 7.) Radiation, quadrupole, and more

Prompt:

46
 d3 x′ T tt (t − r, x′ )x′i x′j
R
Show that the spacial part of the fluctuation h̄ij is lij (t − r) ≡
Solution.
Start with the first thing given,
 
αβ 1 αβ µν
□ h − η h ηµν = 0
2
where,
□ = −∂t2 + ∇2
check this with hµν (t − z), which is only in the z-direction:

□ = −∂t2 + ∂z2

so,
□hµν(t−z) = −∂t2 hij + ∂z2 hij
(WHY IS THIS???) Set to zero,

−∂t2 hij + ∂z2 hij = 0

∂t2 hij = ∂z2 hij

(WHY ALWAYS DOES) This implies,

□hµν (t − z) = 0

The ”0” implies pure vacuum. Use the other equation given:

□h̄αβ = −16πT αβ

If T αβ ̸= 0, use an Einstein equation soluiton:

T αβ (t − |x − x′ |, x′ )
Z
h̄ = 4 d3 x
αβ
|x − x′ |

47