17
$\begingroup$

For any group $G$, cohomology can be viewed as a functor $$ H^\ast(G,-): G{\sf\text{-}mod}\to {\sf GrAbGrp}, $$ where $G{\sf\text{-}mod}$ denotes the category of (left) $\mathbb{Z}[G]$-modules and ${\sf GrAbGrp}$ denotes the category of (non-negatively) graded abelian groups.

It seems that there are non-isomorphic groups $G_1$ and $G_2$ whose integral group rings $\mathbb{Z}[G_1]$ and $\mathbb{Z}[G_2]$ are Morita equivalent (this is a result of Roggenkamp and Zimmermann). One could then ask the following question:

Do there exist non-isomorphic groups $G_1$ and $G_2$ for which there is an equivalence of categories $F: G_1{\sf\text{-}mod}\to G_2{\sf\text{-}mod}$ such that the functors $H^\ast(G_1,-)$ and $H^\ast(G_2,-)\circ F$ from $G_1{\sf\text{-}mod}$ to ${\sf GrAbGrp}$ are naturally isomorphic?

This is my (possibly naive) attempt to formalise the question in the title. It may be that the answer is trivially "no" by looking at $H^0$, ie the functor of coinvariants. In that case, I would like to know if there are other formulations for which the question becomes interesting.

$\endgroup$
1
  • $\begingroup$ A minor formatting suggestion: I would recommend replacing {\sf-mod} with {\sf\text{-}mod}. Even when using the sans-serif font, the symbol - in math mode is interpreted as a minus sign. Compare $G{\sf-mod}$ and $G{\sf\text{-}mod}$. $\endgroup$ Nov 6, 2013 at 14:32

1 Answer 1

15
$\begingroup$

The answer would seem to be yes if I understood your question properly. The paper http://www.m-hikari.com/ija/ija-password-2009/ija-password9-12-2009/ladraIJA9-12-2009.pdf shows that if $\mathbb ZG$ is isomorphic to $\mathbb ZH$, then one can choose an isomorphism which is augmentation preserving. They then show that if $f\colon \mathbb ZG\to \mathbb ZH$ is the augmentation-preserving isomorphism, then viewing a $\mathbb ZH$-module as a $\mathbb ZG$-module gives isomorphisms in homology and cohomology. I didn't check if their proof gives naturality of the isomorphisms, but I would be surprised if it didn't.

The paper http://www.jstor.org/stable/3062112 shows there are nonisomorphic finite groups with isomorphic integral group rings.

$\endgroup$
4
  • $\begingroup$ In fact, it appears that the example in your second link (Martin Hertweck's article "A counterexample to the isomorphism problem for integral group rings") already gives an isomorphism which commutes with the augmentation. I think the relevant statement there is the definition of a group basis in the second phrase of the article. $\endgroup$ Nov 6, 2013 at 15:33
  • $\begingroup$ @RicardoAndrade, thanks. I hadn't read Hertweck's article. I had heard a talk on integral group rings over a year ago in Brazil stating what the example is and was convinced that this should solve the OP's question. So I did a quick search to see if isomorphism of group rings implies that the augmentation must be preserved and found the first link. $\endgroup$ Nov 6, 2013 at 15:36
  • $\begingroup$ Thanks, Benjamin. This seems to answer my question. It now seems obvious that the integral group ring completely determines cohomology. I wasn't aware of this counter-example to the isomorphism problem for integral group rings finite groups (but should have been). By the way, do you know the status of this problem for torsion-free groups? $\endgroup$
    – Mark Grant
    Nov 6, 2013 at 15:46
  • $\begingroup$ @MarkGrant, I'm pretty sure the status of this question was addressed at the talk I attended in Brazil, but I cannot remember what the answer was. Group rings is not really my area but the counterexample and the history leading up to it caught my attention. $\endgroup$ Nov 6, 2013 at 15:58

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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