Let $G$ be a (possibly infinite) group. Let $\mathbb{Z}[G]$ be its integral group ring and let $P$ be a finitely generated projective module over $\mathbb{Z}[G]$. Suppose that the coinvariants of $P$ vanish, i.e. $$ P_G=P/I_G P=0 $$

where $I_G$ is the kernel of the augmentation map $\mathbb{Z}[G]\to\mathbb{Z}$. Does it follow that $P=0$?


This is true when $G$ is finite. This follows from a theorem of Swan [1], which asserts that all projective $\mathbb{Z}[G]$-modules are locally free, i.e isomorphic to a free module over $\mathbb{Z}_p[G]$ for every prime $p$. And clearly, locally free modules have non-trivial coinvariants.

I don't know what happens for infinite groups.

[1] R. G. Swan, Induced representations and projective modules, Ann. of Math. 71 (1960), 552-578.

  • $\begingroup$ This is good news, but actually I am interested in the infinite (finitely presented) case. $\endgroup$ – KotelKanim Jul 23 '15 at 20:31
  • $\begingroup$ Presumably your argument also establishes the case when $G$ is residually finite. Of course there are finitely presented, infinite, simple groups . . . $\endgroup$ – Jason Starr Jul 24 '15 at 13:33

This isn't an area that I'm expert on, and it's quite possible there's a much more elementary and/or more general answer.

But if the Bass Conjecture on Hattori-Stallings ranks for group rings is true for $G$ then I think the answer is yes. I don't think the conjecture has been proved for finitely presented groups in general, but there are large classes of groups for which it is known to be true.

If $A$ is a ring, and $P$ a finitely generated projective $A$-module, then $P$ is the image of an idempotent endomorphism of a finitely generated free module, given by a square matrix $M$ over $A$, and the Hattori-Stallings rank of $P$ is the image in $A/[A,A]$ of the trace of $M$, which depends only on $P$ and not the choice of $M$.

If $A=\mathbb{Z}[G]$ for a group $G$, then $A/[A,A]$ is a free abelian group on the images of a set of conjugacy class representatives $[g]$, and Bass conjectured that the Hattori-Stallings rank of any finitely generated projective is a multiple of $[1]$. Previously (I think), Kaplansky proved that the coefficient of $[1]$ in the rank is strictly positive if $P\not\cong0$, which, if we also assume the Bass conjecture for $G$, proves that the image of $M$ under the augmentation map must have non-zero trace, so that the image of $M$ under the augmentation map is non-zero, which is equivalent to $P$ having a non-zero map to the trivial module $\mathbb{Z}$.

  • $\begingroup$ Yes, it follows from the conjecture of Bass which is open in general, but isn't it still possible that the answer to my question is known nevertheless? $\endgroup$ – KotelKanim Jul 29 '15 at 9:25
  • $\begingroup$ @KotelKanim Yes, that's pretty much what I said in the first paragraph of my answer. $\endgroup$ – Jeremy Rickard Jul 29 '15 at 9:30

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.