Geodesic Deviation

More on geodesic deviation
During one of the Zoom lectures, I was asked about a step in the calculation of geodesic deviation in Lecture
11, particularly the boundary condition on the derivative of the separation vector shown at the bottom of
page 4 of the notes for this lecture. As I examined how to answer this question, I realized that the whole
thing could use a bit of cleanup. My intention in the original presentation of these notes was for certain
terms to be written down exactly, but then for the bulk of the calculation to be done in a local Lorentz
frame (LLF) centered on a particular event.
These notes revisit this entire calculation. I first present geodesic deviation in a very formal way that requires
no particular assumptions about the frame in which the calculation is done. I then revisit the “centered
on the LLF” calculation, re-doing the calculation in Lecture 11. The formal calculation is entirely rigorous,
making no approximations or particular assumptions. The second calculation should build in more physical
insight. This version of the calculation also demonstrates how one can do a computation in a particular
reference frame, but then express it in a tensorial way. The principle that “tensorial in one frame means
tensorial in all frames” then allows us to infer that the resulting expression is good in general. (Plus, in this
case, what we find this way is identical to what we find with the formal calculation.)
First, the formal calculation. Consider a two-parameter family of geodesic trajectories in spacetime xα (λ, s).
The parameter λ is affine parameter along each geodesic; hence uα (s) = (∂xα /∂λ)s is the tangent along the
geodesic with parameter s. (If we choose λ to be proper time, then uα is the usual 4-velocity.)
The parameter s tells us about the separation of neighboring geodesics. The vector1 Y α (λ) = (∂xα /∂s)λ
points from the event at affine parameter λ on the geodesic at s to the event at affine parameter λ on the
geodesic at s + ds.
Note that
∂ 2 xµ ∂ 2 xµ
u β ∇β Y µ = + Γµ βν uβ Y ν and Y β ∇β uµ = + Γµ βν Y β uν .
∂λ∂s ∂s∂λ
Partial derivatives commute, the connection is symmetric on downstairs indices 2 and 3, and we can relabel
dummy indices, so we find uβ ∇β Y µ = Y β ∇β uµ .
Let us now evaluate the covariant acceleration of the separation vector Y µ :
D2 Y µ
uα ∇α uβ ∇β Y µ

≡
dλ2
uα ∇α Y β ∇β uµ

=
uα ∇α Y β (∇β uµ ) + uα Y β ∇α ∇β uµ

=
uα ∇α Y β (∇β uµ ) + uα Y β (∇β ∇α uµ + Rµ ναβ uν )

=
uα ∇α Y β (∇β uµ ) + Y β ∇β (uα ∇α uµ ) − Y β ∇β uα (∇α uµ ) + Rµ ναβ uν uα Y β

=
= Rµ ναβ uν uα Y β .
The first line is essentially a definition. To get to the second line, we use the identity shown above. To
go to the third line, we expand the action of ∇α . To go to the fourth line, we use the commutator rule
[∇α , ∇β ]uµ = Rµ ναβ uν .
In the next step, we use uα (∇β Wα µ ) = ∇β (uα Wα µ ) − (∇β uα ) Wα µ (with Wα µ = ∇α uµ ). To simplify the
fifth line, we use the fact that the first and third terms on the right-hand side are equal but opposite in sign
after using the identity and relabeling dummy indices, as well as the fact that the second term vanishes since
uµ is the tangent along a geodesic trajectory.
The final line gives us the covariant acceleration of the separation vector, and is the equation of geodesic
deviation.
1 This was labeled ξ α in Lecture 11. I am changing here because ξ ~ also denotes a Killing vector, and ξ α used in the lecture
on linearized gravity to describe the generator of an infinitesimal coordinate transformation. The letter ξ is a bit overloaded!
Next, a calculation that focuses on the LLF. Consider two nearby geodesics, both parameterized by affine
parameter λ: Geodesic 1 follows the curve xα (λ), and has tangent vector uα = dxα /dλ; geodesic 2 follows
the curve z α (λ), and has tangent vector v α = dz α /dλ. Let Y α = z α − xα be the vector that points from
events at parameter λ on geodesic 1 to events at λ on geodesic 2.
We expand all important elements of the geometry in a LLF centered on event A at λ = λ0 on geodesic
1. Let A0 be the corresponding event on geodesic 2. We assume that A and A0 are sufficiently close to one
another that their tangent vectors are parallel at λ = λ0 ; we elaborate on what this means quantitatively
below. Other important quantities we need are the metric and the connection at these two events:
Γµ αβ A = 0 , Γµ αβ A0 = Y γ (∂γ Γµ αβ )A .

gµν A = ηµν , gµν A0 = ηµν ;
The geodesic equation for geodesic 1, evaluated at event A, is

d2 xµ

=0.
dλ2 A
The corresponding equation for geodesic 2, evaluated at event A0 , is
d2 z µ

+ Γµ αβ v α v β A0 = 0 .

2
dλ A0
Combining different definitions and assumptions, this can be rewritten

d2 z µ

= −Y γ ∂γ Γµ αβ uα uβ A .

(1)
dλ2 A0
Here we used v α = uα , which follows from our requirement that the tangents to geodesics 1 and 2 be parallel
at λ = λ0 . This requires us to parallel transport from A0 to A, but this transport is trivial2 in the LLF.
Taking the difference between the geodesic equation on geodesic 2 and on geodesic 1, we find
d2 Y µ
= −∂γ Γµ αβ uα uβ Y γ .
dλ2
We drop the A and A0 subscripts from here on; all quantities are now evaluated at A.
This acceleration equation uses d/dλ = uα ∂α as its derivative operator. To write this in a way that nicely
translates into proper tensorial form, we need to express things in terms of D/dλ ≡ uα ∇α . Begin by looking
at the “velocity” equation for the separation vector:
DY µ dY µ
= uα ∇α Y µ = + Γµ αβ Y α uβ .
dλ dλ
You may be tempted to set the connection to zero since we are in the LLF. Wait until we have taken our
next derivative before doing that — the fact that the connection has non-zero slope in the LLF is important.
D2 Y µ
µ
γ dY µ α β
= u ∇γ + Γ αβ Y u
dλ2 dλ
d2 Y µ γ µ dY ν
+ (uγ ∇γ Γµ αβ ) uβ Y α + Γµ γν uγ ∇γ uβ Y α + Γµ αβ uβ (uγ ∇γ Y µ )

= 2
+ u Γ γν
dλ dλ
d2 Y µ
= + ∂γ Γ αβ uβ uγ Y α + O(Γ2 ) .
µ
dλ2
On the final line of this equation, we have gone into the LLF near event A, we have used the fact that uα is
a geodesic, and we used the fact that dY µ /dλ = v µ − uµ = 0.
Finally, use the previous results we found for d2 Y µ /dλ2 to find
D2 Y µ
= ∂γ Γµ αβ uβ uγ Y α − ∂γ Γµ αβ uα uβ Y γ
dλ2
= (∂γ Γµ αβ − ∂α Γµ βγ ) uβ uγ Y α
= Rµ βγα uβ uγ Y α .
2 At least to linear order in important quantities; if you do this transport and keep all terms, you will find an additional term
of order Γ2 . We discard this term at the end of the calculation by our assumption that we are in the LLF.
In going from the first to the second line, we relabeled dummy indices; to go to the third line, we identified
that combination of connection derivatives as the Riemann curvature in the LLF. The final result is identical
to that found in the formal calculation, modulo a few different choices of dummy index labels.
This second version of the calculation highlights how it is that the second order terms in the metric (via
the first order correction to the connection) cause nearby geodesics to deviate from one another. This is the
hallmark of a tidal effect: the geodesics deviate because “free fall” means slightly different things to freely
falling observers that are slightly separated from one another.

Geodesic Deviation

Uploaded by

Copyright:

Available Formats

Geodesic Deviation

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Geodesic Deviation

Uploaded by

Copyright:

Available Formats

More on geodesic deviation

The geodesic equation for geodesic 1, evaluated at event A, is

Combining different definitions and assumptions, this can be rewritten

You might also like