0
$\begingroup$

Definition: Let $\pi:V \rightarrow M$ be a vector bundle. $\Omega^p_{cv}(V)$ is the sections of $p$ form on $V$, such that $\pi^{-1}(K) \cap supp \, (w) $ for all $K \subseteq M$ compact.

Definition: $\Omega^p_c(V)$ is the set of sections $p$ forms on $V$ such that its support is compact.


It is claimed that

$\Omega^p_c(V) = \Omega^p_{cv}(V)$ when $M$ is compact.

How is this true? $\subseteq $ is clear.


Source, Pg 84, Proof of Theorem 6.2.10, line +3

$\endgroup$
2
$\begingroup$

Let $\omega$ be a form on $V$ which is compactly supported on each fiber $V_x$. Consider $\text{supp}(\omega) \subset V$; this is a closed subset of $V$ whose intersection with each fiber is compact. The goal is to show all such sets are compact themselves.

Put a Riemannian metric over V (just so that I may measure "size of vectors"). Then if $X \subset V$ is closed, with the property that each $X \cap V_x$ is compact, then we may define a function $\sigma: M \to \Bbb R_{\geq 0}$ given by sending $\sigma(x)$ to the largest $\|v\|$ of any $v \in X \cap V_x$. This is well-defined because $X \cap V_x$ is compact. The map $\sigma$ may be seen to be continuous by brute-force; I do not want to do it here. (Intuitively, the nearby sets $X \cap V_x$ are 'close', and hence should have 'nearby' size bounds.)

Thus $\sigma$ is a continuous map from a compact space to $\Bbb R$, and hence has an absolute maximum, say $L$. Thus the set $X$ is contained in the closed $L$-disc bundle of $V$, which is a compact space (given any sequence $(m_n, v_n)$ of norm bounded by $L$, choose a subsequence so that $m_n$ is convergent to $m$; working on a chart to identify all nearby fibers with $V_x$, use compactness of the $L$-disc ball to see that a further subsequence may be chosen so that $v_n$ converges). $X$, being a closed subspace of a compact Hausdorff space, is compact.

So any form with compact vertical support on a vector bundle $V$ over compact base $M$ has compact support, as desired.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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