5
$\begingroup$

I am reading John Lee's book and there is a problem on De Rham cohomology: show that the compactly supported De Rham cohomology groups $H^{p}_{DR}(\mathbb{R^n})$ are all zero for $0\leq p<n$. My confusion here lies what is the relationship between compactly supported cohomology and "normal" De Rham cohomology. Since $H^{n}(M)\neq0$, so I can't just use the fact that $\mathbb{R^n}$ is contractible?

$\endgroup$
  • 1
    $\begingroup$ You can't use the fact that $\mathbb R^n$ is contractible since compactly supported cohomology is only a homotopy invariant for homotopies that are proper maps. $\endgroup$ – Eric O. Korman Apr 26 '14 at 16:23
  • $\begingroup$ Then how should I tackle this problem? $\endgroup$ – Alex Apr 26 '14 at 16:26
  • $\begingroup$ Does Lee talk about the long exact sequence in compactly supported cohomology associated to an open (or closed) subset? $\endgroup$ – Eric O. Korman Apr 26 '14 at 17:47
  • $\begingroup$ You mean the M-V sequence? Yes, he did! $\endgroup$ – Alex Apr 26 '14 at 19:10
  • 2
    $\begingroup$ I meant the long exact sequence associated to a pair (see e.g. the l.e.s. at en.wikipedia.org/wiki/Cohomology_with_compact_support with $X, U$ and $Z$). Because you should be able to use that with $X = S^n, Z = pt$ and $U = X - Z \simeq \mathbb R^n$ to get your desired result. $\endgroup$ – Eric O. Korman Apr 26 '14 at 22:13
3
$\begingroup$

Compactly supported cohomology is the cohomology of the complex of compactly supported differential forms (i.e., smooth forms that are zero outside of some compact set). Let's denote compactly supported cohomology and the usual de Rham cohomology by $H^k_c(M)$ and $H^k_{DR}(M)$, respectively.

Note that on compact manifolds, every differential form is compactly supported, so the two notions of cohomology are the same. But on non-compact manifolds, they can be very different. For example, $H^0_{DR}(R^n)$ is generated by the (cohomology class) of the constant function 1. But 1 is not compactly supported, so it isn't allowed in compactly supported cohomology. In fact, $H^0_c(R^n) = 0$. (As a commenter above says, another difference between the two is that compactly supported cohomology is not a homotopy invariant in general.)

An important relationship between the two on orientable manifolds $M$ is given by Poincare duality, which says that $$H^k_{DR}(M) \cong (H^{n-k}_c (M))^\ast$$ where $n$ is the dimension of $M$ and $\ast$ denotes the vector space dual. In particular this means that (at least if everything is finite-dimensional) $H^k_{DR}(M) \cong H^{n-k}_c (M)$. So compactly supported cohomology is the same as de Rham cohomology, but "flipped."

For example, $H^n_c(R^n) \cong H^0_{DR}(R^n) \cong R$, and $H^k_c(R^n) = 0$ for $k < n$. One could probably also compute this directly from the definition of compactly supported cohomology without using Poincare duality (which may be more in the spirit of your problem).

$\endgroup$
  • 2
    $\begingroup$ The way Lee proves poincare duality uses this fact about compactly supported De Rham cohomology, so I wonder if there is a way to tackle this problem directly? $\endgroup$ – Alex Apr 26 '14 at 16:54
  • $\begingroup$ Yes, one could tackle it directly. Does Lee not do this? I have forgotten the details of the argument, but one approach would be an inductive argument, proving that $H^{k}_c(M \times R) \cong H^{k-1}_c(M)$. $\endgroup$ – Phillip Andreae Apr 26 '14 at 17:10

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.