Suppose $G$ is a discrete group given by finitely many generators with finitely many relations. Can the homology groups $H_i(G, \mathbb{Q})$, or equivalently $H_i(BG, \mathbb{Q})$ (topological homology of the classifying space) be infinite-dimensional? Can they be nonzero for infinitely many $i$?

For any finitely presented groups I've seen, the answer is a surprising "no" (all finitely presented groups I know act on a finite-dimensional contractible space with finite stabilizers, and it follows that above the dimension of this space, homology vanishes). But it really should be the case that a "general" finitely-presented group has infinite homology... does anyone know of an example?

  • 1
    $\begingroup$ Just to make sure: the assertion that $H_i(G,\mathbb{Q})$ is finite-dimensional (for all $i>2$) is an assumption, yes? $\endgroup$
    – Alex Suciu
    Aug 7, 2014 at 4:24
  • $\begingroup$ Also, "infinite homology" means "infinite-dimensional homology" (as $\mathbb{Q}$-vector space), right? $\endgroup$
    – Alex Suciu
    Aug 7, 2014 at 4:26
  • $\begingroup$ @Alex: I thought any group homology of a finitely presented group is finite-dimensional. Are there examples where $H_2$ isn't? And yes, "infinite" means infinite-dimensional $\endgroup$ Aug 7, 2014 at 8:07
  • 10
    $\begingroup$ Yes, $H_2$ of an fp group is finitely generated. But, say, $H_3$ needs not be finitely generated. The first such example was given by John Stallings, in a seminal paper, titled, sure enough, A finitely presented group whose 3-dimensional integral homology is not finitely generated, see here. $\endgroup$
    – Alex Suciu
    Aug 7, 2014 at 8:27
  • $\begingroup$ @Dmitry: you should edit the second sentence of your question according to Alex' comments. $\endgroup$
    – YCor
    Aug 7, 2014 at 9:54

2 Answers 2


Thompson's group F is an example. It's finitely presented and, according to this paper of Ken Brown, the integral homology is free abelian of rank 2 in every positive dimension.


This answers the other part of your question, not answered by Thompson's group. For each $i\geq 3$ there is a finitely presented group $G_i$ with the property that $H_i(G_i\mathbb{Q})$ is infinite dimensional. The first such examples were due to Stallings ($i=3$ and Bieri $i>3$). Take a direct product of $i$ copies of the free group $F$ on two generators, and define $G_i$ to be the kernel of the homomorphism $F^i\rightarrow \mathbb{Z}$ that sends each of the $2i$ standard generators to $1\in \mathbb{Z}$. These are the Bieri-Stallings groups. For $i=2$ the analogous group is finitely generated but not finitely presented, and for $i=1$ it is free of infinite rank. These groups can be viewed as special cases of the Bestvina-Brady construction.


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.