It is well known that for two manifolds $M$ and $N$ (let's say they are compact), the Kunneth formula says that $H(M\times N)=H(M)\otimes H(N)$, where $H$ denotes de Rham cohomology with complex coefficients. We also know that one can twist de Rham cohomology by a complex flat vector bundle $E\to M$ with a flat connection $\nabla^E$. Then the differential of the chain complex $\Omega(M; E)$ is coupled with $\nabla^E$. Denote by $H(M, E)$ the corresponding de Rham cohomology.

Let $E\to M$ and $F\to N$ be complex flat vector bundles with flat connections $\nabla^E$ and $\nabla^F$. Write $p_M:M\times N\to M$ be the standard projection map and similarly for $p_N$. Then $p_M^*E\otimes p_N^*F\to M\times N$ is a complex flat vector bundle with flat connection $p_M^*\nabla^E\otimes p_N^*\nabla^F$. I wonder there should be a Kunneth formula for de Rham cohomology twisted by complex flat vector bundles, that is,

$$H(M\times N, p_M^*E\otimes p_N^*F)\cong H(M, E)\otimes H(N, F).$$

Is this assertion obviously true (or obviously wrong)? If the former case is true, is there a reference for a proof of this?


1 Answer 1


Yes, the Künneth formula holds in this framework. Notice that the De Rham cohomology of a flat vector bundle on a manifold is simply the cohomology of the corresponding locally constant sheaf (= local system). So what you are asking for is a Künneth formula for cohomology with local coefficients; this is worked out thoroughly in this paper.

  • $\begingroup$ Thanks for your answer. What about the fiberwise version? Does the Kunneth formula still true? $\endgroup$
    – Ho Man-Ho
    Jul 2, 2016 at 17:57
  • $\begingroup$ Well, this is sheaf cohomology, so there is a Serre spectral sequence as in the case of local coefficients. $\endgroup$
    – abx
    Jul 3, 2016 at 4:25
  • $\begingroup$ @abx: I am intimidated by the algebra in the paper. Denoting by $\mathcal F_{\rho}$ the rank 1 local system with monodromy $\rho: \pi_1(X) \to \mathbb C^*$, does theorem 1.6 (or 1.7) imply that $H^k(X\times Y, \mathcal F_{(\rho_1, \rho_2)}) = \sum_{p+q=k} H^p(X, \mathcal F_{\rho_1}) \otimes H^q(X, \mathcal F_{\rho_2})$ if the spaces $X,Y$ are sufficiently well behaved? $\endgroup$
    – klirk
    Aug 27, 2019 at 20:33

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.