MathOverflow is a question and answer site for professional mathematicians. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

Hi there.

Assume $(M,g)$ is a Riemanian manifold and $E\to M$ is a vector bundle with a bundle metric $\langle\cdot,\cdot\rangle$. We then have the pre-Hilbert space $H_0:=\Gamma_c^\infty(E)$ of compactly supported smooth sections with $(s_1,s_2):=\int_M\langle s_1,s_2\rangle dV_g$ . In a paper i'm currently working on it says that the completion $H$ of $H_0$ is (in a rather concrete Situation for $E$) given by the space of square integrable sections $L^2(M,E):=\{s\in\Gamma(E)\mid \int_M|\langle s,s\rangle|^2dV_g<\infty\}$, identifying sections being equal almost everywhere. Is this true in general and if yes, do you know a book where this is worked out?

Greetings, Robert Rauch

share|cite|improve this question
This seems more appropriate for than for MathOverflow. I also suggest that you try to work out the details yourself. It's really the same as how you define $L^2(R)$ as the completion of compactly supported smooth functions on $R$. – Deane Yang Nov 30 '11 at 4:59
up vote 2 down vote accepted

This is true in general. I don't know a reference for the statement, but it is pretty simple just to work it out. The point is that $L^2(M,E)$ is a Hilbert space which contains $H_0$ as a dense linear subspace, so it must be the completion.

share|cite|improve this answer
I guess I should add that $H_0$ sits isometrically inside $L^2(M,E)$. – MTS Nov 29 '11 at 22:01
What you are essentially saying is, that i should check the definition of the completition. The point is, its not that obvious to me that $L^2(M,E)$ is a Hilbertspace. I mean, is there a simpler argument than "go along the proof for the $L^p(X,\mu)$ spaces and change everything such that it fits our needs"? – Robert Rauch Nov 29 '11 at 22:20
Well, essentially one does need to develop a theory of integration of Banach space-valued functions. This theory is developed, for example, in some lecture notes (from 1970) by Marc Rieffel, posted here: – MTS Nov 29 '11 at 22:50
I also thought about this in the beginning, but we are not going to define an integral for sections in $E$ (in which space should $\int_M s dV$ lie for $s\in\Gamma(E)$?!), but a bilinear form $(\cdot,\cdot)$ on a certain subspace of $\Gamma(E)$, which involves only integration of, say $\mathbb{C}$-valued functions on the measure space $(M,dV)$. – Robert Rauch Nov 29 '11 at 23:34
I do not understand why you should need Banach-space-valued functions. The pointwise norm of a section defines an ordinary function on M, in particular it is clear what to integrate. – Orbicular Nov 30 '11 at 0:02

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.