Sectional curvature: Difference between revisions

From Diffgeom
m (9 revisions)
 
Line 54: Line 54:
''This is analogous to the fact that [[Ricci curvature]] determines the [[Ricci curvature tensor]]''
''This is analogous to the fact that [[Ricci curvature]] determines the [[Ricci curvature tensor]]''


Given the sectional curvature for all tangent planes, we can back-calculate the Riemann curvature tensor from it by the usual polarization trick. The idea is to view the Riemann curvature tensor a s a symmetric bilinear form on <math>\Lambda^2(TM)</math>, and the sectional curvature as the values that this form takes on pairs <math>(\Pi,\Pi)</math> of ''unit vectors'' in <math>\Lambda^2(\Gamma(TM))</math>. To clarify:
The following facts are true about the Riemann curvature tensor:


* The [[Riemann curvature tensor of Levi-Civita connection|Riemann curvature tensor]] can be viewed as a <math>(0,4)</math>-tensor:
* The [[Riemann curvature tensor of Levi-Civita connection|Riemann curvature tensor]] can be viewed as a <math>(0,4)</math>-tensor:
Line 62: Line 62:
* The above map is antisymmetric in <math>X, Y</math> ({{proofat|[[Antisymmetry of Riemann curvature tensor]]}} and antisymmetric in <math>Z</math> and <math>W</math>. Further it is symmetric in the pairs <math>(X,Y)</math> and <math>(W,Z)</math>. Hence, it is a symmetric bilinear form on <math>\Lambda^2(\Gamma(TM))</math>.
* The above map is antisymmetric in <math>X, Y</math> ({{proofat|[[Antisymmetry of Riemann curvature tensor]]}} and antisymmetric in <math>Z</math> and <math>W</math>. Further it is symmetric in the pairs <math>(X,Y)</math> and <math>(W,Z)</math>. Hence, it is a symmetric bilinear form on <math>\Lambda^2(\Gamma(TM))</math>.


* A unit vector in <math>\Lambda^2(\Gamma(TM))</math> defines on each tangent space, an element of <math>\Lambda^2(T_pM)</math> that has norm 1. The norm of an element in the second exterior power is the square of the area of the parallelogram it spans, so unit norm vectors are things of the form <math>X \wedge Y</math> where <math>X,Y</math> span a parallelogram of area <math>1</math>. Thus, the denominator vanishes for these pairs, and the sectional curvature is simply <math>R(X,Y,Y,X)</math> which is the inner product of <math>X \wedge Y</math> with itself.
Thus, given the sectional curvature for all tangent planes, we can back-calculate the Riemann curvature tensor from it by the usual polarization trick. The idea is to view the Riemann curvature tensor as a symmetric bilinear form on <math>\Lambda^2(TM)</math>, and the sectional curvature as the values that this form takes on pairs <math>(\Pi,\Pi)</math> of (special kinds of) ''unit vectors'' in <math>\Lambda^2(\Gamma(TM))</math>.

Latest revision as of 12:38, 22 May 2008

This article defines a notion of curvature for a differential manifold equipped with a Riemannian metric

The equivalent notion for a pseudo-Riemannian manifold is: Sectional curvature for a pseudo-Riemannian manifold

Definition

Given data

A Riemannian manifold viz a differential manifold equipped with a Riemannian metric .

Definition part

Let be a tangent plane to at a point . Then, the sectional curvature of at is defined as follows: take two linearly independent vectors and in , and calculate:

viz the inner product of and with respect to .

Divide this by the square of the area of the parallelogram formed by and . This ratio defines the sectional curvature of , denoted as .

Here is the more explicit formula:

Note that the denominator cannot vanish because and are independent vectors.

For a pseudo-Riemannian manifold

Further information: Sectional curvature for a pseudo-Riemannian manifold

We can also define the sectional curvature of a pseudo-Riemannian manifold. The same definition works.

Related notions

Related notions of curvature

Related metric properties

Facts

Sectional curvature determines Riemann curvature tensor

This is analogous to the fact that Ricci curvature determines the Ricci curvature tensor

The following facts are true about the Riemann curvature tensor:

  • The above map is antisymmetric in (For full proof, refer: Antisymmetry of Riemann curvature tensor and antisymmetric in and . Further it is symmetric in the pairs and . Hence, it is a symmetric bilinear form on .

Thus, given the sectional curvature for all tangent planes, we can back-calculate the Riemann curvature tensor from it by the usual polarization trick. The idea is to view the Riemann curvature tensor as a symmetric bilinear form on , and the sectional curvature as the values that this form takes on pairs of (special kinds of) unit vectors in .