12

If $\mathcal{M}$ is a Riemannian manifold of constant curvature, is the manifold $\mathcal{M}^n$ with the product metric, of constant curvature? (and why?)

Thank you

Chevallier
  • 1,062
  • I think that thw scalar curvature of a product manifold is the sum of the scalar curvatures of its factors. But you should double check on some textbook on Riemannian geometry. – Giuseppe Negro Dec 30 '15 at 21:47

1 Answers1

14

For your question, the answer is

Proposition 1: Let M be a Riemannian manifold with constant curvature, then $M^n$ has constant curvature if and only if $M$ has constant curvature zero.

For example, the $S^2\times S^2$ is not a constant curvature space as the 2 dimensional plain spanned by two vectors which come from the tangent spaces of two $S^2$ has sectional curvature zero.

Let's prove a more general result here.

Proposition 2: Let $M=M_1\times M_2$ be the product of two riemannian manifolds, and R be its curvature tensor, $R_1, R_2$ be curvature tensor for $M_1$ and $M_2$ respectively, then one can relate $R, R_1$ and $R_2$ by

$$R(X_1+X_2,Y_1+Y_2,Z_1+Z_2,W_1+W_2)=R_1(X_1,Y_1,Z_1,W_1)+R_2(X_2,Y_2,Z_2,W_2)$$

where $X_i, Y_i, Z_i, W_i\in TM_i$.

To show this, you should use:

(1)$\langle X_1+X_2,Y_1+Y_2 \rangle_{M}=\langle X_1,Y_1 \rangle_{M_1}+\langle X_2,Y_2 \rangle_{M_2}$;

(2)$[X_1+X_2,Y_1+Y_2]_{M}=[X_1,Y_1]_{M_1}+[X_2,Y_2]_{M_2}$;

(3)$\nabla_{X_1+X_2}^M(Y_1+Y_2)=\nabla_{X_1}^{M1}(Y_1)+\nabla_{Y_1}^{M2}(Y_2)$.

(1) is simply by definition of product Riemannian manifold, (2) can be shown in local coordinates, and (3) can be shown by (1) and (2) and along with Koszul formula. Also, you may find this post useful.

AG learner
  • 4,523
  • I don't see how Proposition 1 follows from Proposition 2. Doesn't the latter show that you can omit the "zero" in the former, since the curvature of the product is the direct sum of the curvatures of the components? – Bananach Oct 04 '21 at 08:06
  • 1
    @Bananach: let $M, M'$ be two copies of the same constant curvature Riemannian manifold, let $v, v'\in T_pM, T_pM'$ respectively. Then $R(v, v', v, v') =0$, – Willie Wong Oct 31 '21 at 13:18
  • @Willie Thanks. If I may spell out the remaining details: "... because $R(v, v', v,v')=R_1(v,0,v,0)+R_1(0,v',0,v')=0+0$. Thus, if the product has constant curvature, then that constant must be zero. But $R_1(v,w,v,w)=R(v,w,v,w)=0$ for any $v,w\in T_pM$ and therefore $M$ has constant zero curvature as well". I was stuck thinking that constant plus constant is clearly always constant, but forgot that constant curvature is defined in terms of sectional curvatures – Bananach Oct 31 '21 at 14:27
  • Some details: http://www.mathem.pub.ro/dgds/v05/D05-ATC.pdf – ABIM Jun 09 '23 at 16:52