# Integrating on Compact Manifolds

1. Mar 22, 2012

### AnalysisQuest

1. The problem statement, all variables and given/known data
This problem is in Analysis on Manifolds by Munkres in section 25. $R$ means the reals
Suppose $M \subset R^m$ and $N \subset R^n$ be compact manifolds and let $f: M \rightarrow R$ and $g: N \rightarrow R$ be continuous functions.

Show that $\int_{M \times N} fg = [\int_M f] [ \int_N g ]$

The hint states to consider the case where supports of f and g are contained in coordinate patches.

2. Relevant equations
I know the general formulation of integration over a compact manifold when the support of the function is contained in a single coordinate patch. Also, I know how to integrate functions on manifolds using partitions of unity.

3. The attempt at a solution
I know the general idea. First I must prove that this holds when the support of f and g are in a single coordinate patch. Then using partitions of unity, I want to notice the integral is just the sum of integrals over finitely many coordinate patches. What I can't get is the part where the support of f and g are in a single coordinate patch:

Suppose $supp(f) \subset V$ with $\phi : U \rightarrow V \subset M$ and $supp(g) \subset V'$ with $\psi : U' \rightarrow V' \subset N$.

Then $\int_{M\times N} fg = \int_{U \times U'} (fg) \circ (\phi\times\psi) V_k(\phi \times \psi)$ but I don't know how to deal with this...

2. Mar 22, 2012

### Dick

I think you just want to use Fubini's theorem. The integral of |fg| exists because f and g are continuous on compact sets, so they are uniformly continuous and bounded. Doesn't that work?

3. Mar 22, 2012

### AnalysisQuest

The problem is that I don't know how to split up the k-volume term and I don't know how to deal with phi x psi. Unless what I've done so far is completely wrong. I've been stuck for days now and I'm just frustrated and missing something obvious, I think.

4. Mar 22, 2012

### Dick

I think you are over complicating it. If you are working on a single coordinate patch and you've applied a partition of unity then the functions you are integrating are continuous and the domains over which you are integrating have compact support. The Jacobean is differentiable. Everything is bounded. I think you can just say it's Fubini's theorem on R^m x R^n. I don't think you need to do anything complicated.

Last edited: Mar 22, 2012