Curvature is antisymmetric in last two variables: Difference between revisions

From Diffgeom
No edit summary
Line 27: Line 27:
<math>g(\nabla_{[X,Y]}(Z),W) + g(\nabla_{[X,Y]}(W),Z) = [X,Y]g(Z,W) = XYg(Z,W) - YXg(Z,W) \qquad (\dagger\dagger)</math>.
<math>g(\nabla_{[X,Y]}(Z),W) + g(\nabla_{[X,Y]}(W),Z) = [X,Y]g(Z,W) = XYg(Z,W) - YXg(Z,W) \qquad (\dagger\dagger)</math>.


Simplifying each of the two terms on the right side of <math>\tag{\dagger\dagger}</math>, we get:
Simplifying each of the two terms on the right side of <math>(\dagger\dagger)</math>, we get:


<math>XYg(Z,W) = Xg(\nabla_Y(Z),W) + Xg(Z,\nabla_Y(W)) = g(\nabla_X \circ \nabla_Y(Z),W) + g(\nabla_Y(Z),\nabla_X(W)) + g(Z,\nabla_X \circ \nabla_Y(W)) + g(\nabla_X(Z),\nabla_Y(W)) \qquad (1)</math>.
<math>XYg(Z,W) = Xg(\nabla_Y(Z),W) + Xg(Z,\nabla_Y(W)) = g(\nabla_X \circ \nabla_Y(Z),W) + g(\nabla_Y(Z),\nabla_X(W)) + g(Z,\nabla_X \circ \nabla_Y(W)) + g(\nabla_X(Z),\nabla_Y(W)) \qquad (1)</math>.

Revision as of 01:50, 24 July 2009

Statement

Suppose is a differential manifold and is a Riemannian metric or pseudo-Riemannian metric and is the Levi-Civita connection for . Consider the Riemann curvature tensor of . In other words, is the Riemann curvature tensor of the Levi-Civita connection for . We can treat as a -tensor:

.

Then:

.

Proof

We consider the expression :

By the bilinearity of , this simplifies to:

To prove that this is zero, it thus suffices to show that:

.

We now show . Since is a metric connection, the left side simplifies to:

.

Simplifying each of the two terms on the right side of , we get:

.

And:

.

Substituting (1) and (2) in yields .