Differential geometry MOC
Smooth geodesic
On a πΆ πΌ -manifold , there are two ways to go about defining a geodesic :
As the straightest path between two points, i.e. tangent vectors are parallel by Parallel transport using the affine connexion ;
As a shortest or extremizing path between two points, i.e. a path of maximal or minimal length as defined using the metric tensor .1
Note that the latter only makes sense for a path which is definite, i.e. the line element is strictly positive or strictly negative.2
When the connexion used is the Levi-Civita connexion , these notions coΓ―ncide.3
Straightest path
Let π be a πΆ πΌ -manifold equipped with an affine connexion β .
Consider a smooth path
πΎ : πΌ β π : π β¦ πΎ ( π )
with Λ πΎ = d πΎ / d π .
We say πΎ is a geodesic iff its tangent vectors are related by parallel transport along πΎ , i.e.
Λ πΎ π β π Λ πΎ π = 0 .
all along πΎ .
In local coΓΆrdinates π₯ : π β π we therefore have
Β¨ π₯ πΌ + Ξ πΌ π π Λ π₯ π Λ π₯ π = 0 .
This is parameterization dependent, since it cares about the speed at which we traverse the curve, and will only consider βaffine parameterizationsβ as straight.
Extremizing path
Let ( π , π π π ) be a semi-Riemannian manifold .
Consider a definite smooth path
πΎ : [ 0 , 1 ] β π : π β¦ πΎ ( π )
and let
Λ π : = d d π π ( πΎ ( π ) )
for any smooth function π : π β π .
The path πΎ induces a 1-dimensional pullback metric on [ 0 , 1 ] so that in local coΓΆrdinates π₯ : π β β π we have
d π 2 = π π π Λ π₯ π Λ π₯ π d π 2
where everything is a function of π .
Since πΎ is definite, without loss of generality we may assume that the factor in front of d π 2 is nonnegative.4
We may thus define the length functional
β [ πΎ ] = β« π β [ 0 , 1 ] d π = β« π β [ 0 , 1 ] β π π π Λ π₯ π Λ π₯ π d π = β« π β [ 0 , 1 ] πΏ π π
where we have introduced the Lagrangian function
πΏ = πΏ ( ( π₯ π ) , ( d π₯ π d π ) ) : = β π π π Λ π₯ π Λ π₯ π .
We wish to find the extermizing path for the functional β .
By the Fundamental theorem of calculus , it is clear that πΏ = d π / d π and thus for π = π ( π ( π ) ) we have
Λ π = d π d π d π d π = d π d π πΏ .
It follows from the Euler-Lagrange equations that
π πΏ π π₯ πΌ β d d π π πΏ π ( d π₯ πΌ / d π ) = 0 .
For the partial derivatives with respect to π₯ πΌ we have
π πΏ π π₯ πΌ = β 1 2 πΏ π π π π π π₯ πΌ Λ π₯ π Λ π₯ π = β πΏ 2 π π π π π π₯ πΌ d π₯ π d π d π₯ π d π .
For the partial derivatives with respect to Λ π₯ πΌ we note
d d Λ π₯ πΌ [ π π π Λ π₯ π Λ π₯ π ] = π π π ( Λ π₯ π πΏ π πΌ + Λ π₯ π πΏ π πΌ ) = 2 π πΌ π Λ π₯ π
by the product rule and thus
π πΏ π Λ π₯ πΌ = β 1 πΏ π πΌ π Λ π₯ πΌ .
Differentiating with respect to π and eliminating instances of πΏ using derivatives with respect to π , we have
β d d π ( π πΏ π Λ π₯ πΌ ) = d d π ( 1 πΏ π πΌ π Λ π₯ πΌ ) = d π π ( π πΌ π d π₯ π d π ) πΏ = ( π πΌ π d 2 π₯ π d π 2 + d π πΌ π d π d π₯ π d π ) πΏ = ( π πΌ π d 2 π₯ π d π 2 + π π πΌ π π π₯ π d π₯ π d π d π₯ π d π ) πΏ = ( π πΌ π π 2 π₯ π π π 2 + 1 2 ( π π πΌ π π π₯ π + π π πΌ π π π₯ π ) d π₯ π d π d π₯ π d π ) πΏ .
Thus the Euler-Lagrange equations say
0 = ( π πΌ π d 2 π₯ π d π 2 + 1 2 ( π π πΌ π π π₯ π + π π πΌ π π π₯ π ) d π₯ π d π d π₯ π d π ) πΏ β πΏ 2 π π π π π π₯ πΌ d π₯ π d π d π₯ π d π .
We divide out by πΏ to get
π πΌ π d 2 π₯ π d π 2 = 1 2 ( π π π π π π₯ πΌ β π π πΌ π π π₯ π β π π πΌ π π π₯ π ) d π₯ π d π d π₯ π d π
and raising indices gives
d 2 π₯ πΌ d π 2 + Ξ πΌ π π d π₯ π d π d π₯ π d π = 0
where the Christoffel symbols are defined by
Ξ πΌ π π = 1 2 ( π π πΌ π π π₯ π + π π πΌ π π π₯ π β π π π π π π₯ πΌ ) .
develop | en | SemBr