Take the 2-minute tour ×
MathOverflow is a question and answer site for professional mathematicians. It's 100% free, no registration required.

For a trivial action on the coefficient, we have the following Kuenneth formula for group cohomology:

$$ H^n(G_1 \times G_2; M) \cong [\oplus_{i= 0}^n H^i(G_1;M) \otimes_M H^{n-i}(G_2;M)] \oplus [\oplus_{p =0}^{n+1} \text{Tor}^M(H^p(G_1;M),H^{n+1-p}(G_2;M))] $$ where $G_1$ and $G_2$ are finite groups and/or compact Lie groups --Edit-- and $M$ is a PID such as $Z$?

If the group action on the coefficient $M$ is non-trivial, does the above Kuenneth formula remain valid?

share|improve this question
    
(I guess $M$ is a ring.) There is not a formula that's quite this simple except in some cases. If $G_1$ and $G_2$ are finite groups, then there is what is called a Tor spectral sequence, which requires contributions from the higher Tor groups over $M$ and is, unfortunately, not quite as simple as all of the Tor groups appearing in a direct sum. If the coefficient actions are not trivial then it gets more complicated. –  Tyler Lawson Sep 15 '11 at 2:14
    
I am sorry. I should mention that $M$ is a PID, such as $Z$. Also, the action of $G_1\times G_2$ may not be most general. For example, we may assume that only $G_1$ acts non-trivially on $M$. In fact, $G_1\times G_2$ acts "naturally", say, on $H^i(G_1,M)\otimes_Z H^{d−i}(G_2,M)$, Let $a\in H^i(G_1,M)$ and $b\in H^{d−i}(G_2,M)$. We have a group action $(g_1,g_2) \cdot (a\otimes b)=(g_1\cdot a)\otimes (g_2\cdot b)$. –  Xiao-Gang Wen Sep 15 '11 at 2:23
2  
There is a Künneth formula but only when the coefficient is a tensor product $A\bigotimes B$ (and one of them is flat over the base ring). For trivial action and $A=B$ is equal to the base ring we have $A\bigotimes B$ is again equal to the base ring with trivial action. In the general case the action may not factor in that way. However, you seem to be content with $G_1$ acting trivially and then you do have such a tensor product factorisation and everything is OK. –  Torsten Ekedahl Sep 15 '11 at 4:44
add comment

1 Answer

up vote 4 down vote accepted

I learned this through Ken Brown's textbook "Cohomology of Groups" (and studying under him): I basically use the beginning of his Chapter 5 and solve the two exercises in that section.

Let $M$ (resp. $M'$) be an arbitrary $G$-module (resp. $G'$-module), let $F$ (resp. $F'$) be a projective resolution of $\mathbb{Z}$ over $\mathbb{Z}G$ (resp. $\mathbb{Z}G'$), and consider the map $(F\otimes_GM)\otimes(F'\otimes_{G'}M')\rightarrow (F\otimes F')\otimes_{G\times G'}(M\otimes M')$ given by $(x\otimes m)\otimes(x'\otimes m')\mapsto(x\otimes x')\otimes(m\otimes m')$.

Note that $(F\otimes_GM)$ $\otimes(F'\otimes_{G'}M')$ = $(F\otimes M)_G \otimes(F'\otimes M')_G$ ,

which is the quotient of $F\otimes M\otimes F'\otimes M'$ by the subgroup generated by elements of the form $gx\otimes gm\otimes g'x'\otimes m'$. The isomorphism $F\otimes M\stackrel{\cong}{\rightarrow}M\otimes F$ of chain complexes ($M$ in dimension $0$) is given by $x\otimes m\mapsto (-1)^{deg(m)\cdot deg(x)}m\otimes x=m\otimes x$, and so the aforementioned quotient is isomorphic to $F\otimes F'\otimes M\otimes M'$ modulo the subgroup generated by elements of the form $(gx\otimes g'x'\otimes gm\otimes g'm')=(g,g')\cdot(x\otimes x'\otimes m\otimes m')$ where this latter action is the diagonal $(G\times G')$-action. Now this is precisely

$$(F\otimes F'\otimes M\otimes M')_{G\times G'} = (F\otimes F')\otimes_{G\times G'}(M\otimes M')$$ and hence the considered map is an isomorphism.

Assuming now that either $M$ or $M'$ is $\mathbb{Z}$-free, we have a corresponding Künneth formula $\bigoplus_{p=0}^nH_p(G,M)\otimes H_{n-p}(G',M')\rightarrow H_n(G\times G',M\otimes M')$

$\rightarrow\bigoplus_{p=0}^{n-1}Tor_1^\mathbb{Z}(H_p(G,M),H_{n-p-1}(G',M'))$ by Proposition I.0.8[Brown]. Note that in order to apply the proposition we needed one of the chain complexes (say, $F\otimes_G M$) to be dimension-wise $\mathbb{Z}$-free (and so with a free resolution $F$ this means we needed $M$ to be $\mathbb{Z}$-free).

Actually, the general Künneth theorem has a more relaxed condition and it suffices to choose $M$ (or $M'$) as a $\mathbb{Z}$-torsion-free module}.


Cohomology Künneth Formula (no proofs, just statements/notes)

Let $M$ (resp. $M'$) be an arbitrary $G$-module (resp. $G'$-module), let $F$ (resp. $F'$) be a projective resolution of $\mathbb{Z}$ over $\mathbb{Z}G$ (resp. $\mathbb{Z}G'$), and consider the cochain cross-product $Hom_G(F,M)\otimes Hom_{G'}(F',M')\rightarrow Hom_{G\times G'}(F\otimes F',M\otimes M')$ which maps the cochains $u$ and $u'$ to $u\times u'$ given by $\langle u\times u',x\otimes x'\rangle=(-1)^{deg(u')\cdot deg(x)}\langle u,x\rangle\otimes\langle u',x'\rangle$. This map is an isomorphism under the hypothesis that either $H_i(G,M)$ or $H_i(G',M')$ is of finite type, that is, the $i$th-homology group is finitely generated for all $i$ (alternatively, we could simply require the projective resolution $F$ or $F'$ to be finitely generated).

For example, if $M=\mathbb{Z}$ then $Hom_G(-,\mathbb{Z})$ commutes with finite direct sums, so we need only consider the case $F=\mathbb{Z}G$. An inverse to the above map is given by $t\mapsto \varepsilon\otimes \phi$, where $\varepsilon$ is the augmentation map and $\phi:F'\rightarrow M'$ is given by $\phi(f')=t(1\otimes f')$. Note that this does not hold for infinitely generated $P=\bigoplus^\infty \mathbb{Z}G$ because $Hom_G(P,\mathbb{Z})\cong\prod^\infty \mathbb{Z}$ is not $\mathbb{Z}$-projective (i.e. free abelian).

Assuming now that either $M$ or $M'$ is $\mathbb{Z}$-free, we have a corresponding Künneth formula $\bigoplus_{p=0}^nH^p(G,M)\otimes H^{n-p}(G',M')\rightarrow H^n(G\times G',M\otimes M')\rightarrow$

$\bigoplus_{p=0}^{n+1}Tor_1^\mathbb{Z}(H^p(G,M),H^{n-p+1}(G',M'))$ by Proposition I.0.8[Brown].

share|improve this answer
    
In particular, since every module over a field $k$ is free, $Tor^k_n(A,B)=0$ for all $n>0$ via the free resolution $0\rightarrow A\rightarrow A\rightarrow 0$, and so the Kunneth formula implies $H^*(G\times H,k)\cong H^*(G,k)\otimes_k H^*(H,k)$. –  Chris Gerig Sep 15 '11 at 5:49
    
@Chris: I fixed some of your tex using backticks, at the expense of making the font smaller. Feel free to change it back. –  Mark Grant Sep 15 '11 at 6:50
    
@ Chris: Thank you very much. I want to make sure that when describing Ku¨nneth formula, do you really mean the exact sequence $0\rightarrow \bigoplus_{p=0}^nH^p(G,M)\otimes H^{n-p}(G',M')\rightarrow H^n(G\times G',M\otimes M')\rightarrow$ $\bigoplus_{p=0}^{n+1}Tor_1^\mathbb{Z}(H^p(G,M),H^{n-p+1}(G',M'))\rightarrow 0$ –  Xiao-Gang Wen Sep 15 '11 at 11:18
    
@Mark: Thanks! @Xiao: Yup! –  Chris Gerig Sep 15 '11 at 20:22
    
@ Chris - Does the above exact sequence for cohomology always split? –  Xiao-Gang Wen Sep 17 '11 at 14:41
show 5 more comments

Your Answer

 
discard

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.