2
$\begingroup$

The question:

(1) Is every compact Lie group $G$ isomorphic (as a topological group) to some quotient $H/N$ where $H$ is a torsion-free compact metrizable group?

Or equivalently:

(2) Is every compact metrizable group $G$ isomorphic (as a topological group) to some quotient $H/N$ where $H$ is a torsion-free compact metrizable group?

((1)$\implies$(2) follows from the fact that every compact metrizable group is isomorphic to an inverse limit of compact Lie groups)

A related question that came up in our research:

(3) Is there a compact connected metrizable non-abelian torsion-free group?

(I asked the latter question without the assumption on connectivity here, and YCor pointed out that the Heisenberg group over the $p$-adic integers is a suitable example.

$\endgroup$

1 Answer 1

3
$\begingroup$

The answer to all questions is "no" (and even without the metrizability requirements).

a) A negative answer to (3) implies a negative answer to (1).

Indeed, if $f:G\to H$ is a surjective continuous homomorphism between compact groups, then $G^\circ\to H^\circ$ is surjective as well. So a non-abelian connected compact group (such as $\mathrm{SO}(3)$) is not quotient of any torsion-free compact group.

b) The answer to (3) is negative: every connected torsion-free compact group $G$ is abelian. Indeed, let $G$ be a connected compact group. Then (see Bourbaki Lie, Chap. IX Appendix I) there exists a family $(S_i)_{i\in I}$ of connected simple Lie groups (with finite center) and a surjective homomorphism $S=\prod S_i\to [G,G]$ whose kernel is central in $S$. Since each $S_i$ contains non-central elements of finite order, it follows that $[G,G]$ is not torsion-free, unless $I$ is empty, in which case $G$ is abelian.

$\endgroup$
4
  • $\begingroup$ (Side remark: if $W$ is the Pontryagin dual of $\mathbf{Q}$, the connected torsion-free compact groups are thus the powers $W^X$ of $W$. Indeed a connected torsion-free compact abelian group is the same as the Pontryagin dual of a discrete torsion-free divisible group, and this is always of the form $\mathbf{Q}^{(X)}$ for some set $X$.) $\endgroup$
    – YCor
    Mar 28, 2022 at 18:19
  • $\begingroup$ Thank you! I understand it except for the reference. I assume you referred to Bourbaki Chap. IX, Appendix I, Sec. 3, Prop. 2/c (in the 2004 edition). It says that for a connected compact group $G$ there exists a family $(S_\lambda)_{\lambda\in L}$ of almost simple compact Lie groups and a surjective cont. hom. $\prod_{\lambda\in L} S_\lambda \to D(G)$, whose kernel is a totally discontinuous, compact, central subgroup. In Chapter III almost simple Lie groups are defined but the definition uses the notion of integral subgroups. I could not find what integral subgroups are. Can you help? $\endgroup$
    – chj
    Mar 30, 2022 at 15:51
  • $\begingroup$ An almost simple compact Lie group means a non-abelian compact Lie group in which every proper normal (closed) subgroup is finite central. (I didn't look up in Bourbaki) $\endgroup$
    – YCor
    Mar 30, 2022 at 16:40
  • $\begingroup$ With this definition everything is clear, thank you very much! $\endgroup$
    – chj
    Mar 31, 2022 at 7:05

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.