It is known (and not complicated to prove) that for a finitely generated not virtually nilpotent group $G$, we can pass to a quotient $G/N$ of $G$, such that the quotient is just not virtually nilpotent, i.e. it is not virtually nilpotent, but every proper quotient of it is.

Does a similar argument holds in the broader realm of locally compact groups? that is,

Does any compactly generated locally compact non nilpotent-by-compact group $G$ have a non nilpotent-by-compact quotient $G/N$ such that any proper quotient of $G/N$ is nilpotent-by-compact?

One may assume some relaxing conditions, for example that $G$ is a real connected Lie group.

  • 1
    $\begingroup$ In the setting of locally compact groups, it is better to consider the class of groups with polynomial growth, because it is more stable. Indeed the class of nilpotent-by-compact groups is not stable under taking closed subgroups. $\endgroup$
    – YCor
    Commented Jul 14, 2015 at 11:37
  • $\begingroup$ In the case of real connected Lie groups, a result in the spirit of what you're looking at is Lemma 4.1 in msp.org/gt/2008/12-1/gt-v12-n1-p11-s.pdf $\endgroup$
    – YCor
    Commented Jul 14, 2015 at 11:54
  • $\begingroup$ Ha ha, I was just going through a paper of yours looking for clues, thanks @YCor! $\endgroup$
    – Snoop Catt
    Commented Jul 15, 2015 at 7:18
  • $\begingroup$ A small follow up question - in theorem 4.3 of msp.org/gt/2008/12-1/gt-v12-n1-p11-s.pdf, you do not assume the the Lie group is real, but you invoke Lemma 4.1, which is proved for this case only. What am I missing? Thanks! $\endgroup$
    – Snoop Catt
    Commented Jul 19, 2015 at 10:10
  • $\begingroup$ In theorem 4.3 it's written "connected Lie group", so there's little ambiguity that it's a real Lie group. A connected $p$-adic Lie group is necessarily reduced to the trivial group, and a complex Lie group can be viewed as a real Lie group. $\endgroup$
    – YCor
    Commented Jul 19, 2015 at 12:30

1 Answer 1


Every compactly generated locally compact group is either of polynomial growth, or it has a quotient that is just non-(polynomial growth). The same also works for a number of similar properties in place of 'polynomial growth', e.g. 'compact-by-nilpotent'. See Theorem 3.12 of this preprint: http://arxiv.org/abs/1509.06593 (it should say "for any compactly generated locally compact group $G$...")

I suspect there are other short proofs of this result using only the published literature, but this is the argument I know of.


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.