12
$\begingroup$

Suppose $G$ is a finite group and $A$ an abelian subgroup. Suppose for some natural number $n\geq 2$, elements of $\gamma_n(G)$ have the form $[a, x]$ where $a\in A$ and $x\in G$. Then $G$ is solvable.

$\endgroup$
3
  • $\begingroup$ A similar statement is true for finite dimensional Lie algebras of characteristic zero. See my short note in Bull. Australian Math. Soc.(2011): A note on derivations of Lie algebras. $\endgroup$
    – Sh.M1972
    Dec 5, 2012 at 17:03
  • 1
    $\begingroup$ and what is $\gamma_n(G)$? $\endgroup$ Dec 5, 2012 at 17:38
  • $\begingroup$ @Dima: it is the $n$-th term of the lower central series. $\endgroup$
    – Sh.M1972
    Dec 5, 2012 at 18:34

2 Answers 2

14
$\begingroup$

I claim that the conjecture is false, and there is a counterexample with $G=S_5$, $n=2$, and $A$ cyclic of order 6. In this case $\gamma_2(G)=[G,G]=G'=A_5$ has order $5!/2=60$. Take $A$ to be the abelian group $A=\langle(1,2,3),(4,5)\rangle$, but any conjugate of $A$ will also work. A simple computation shows that the set $C:=\{[a,x]\mid a\in A,\ x \in G\}$ also has 60 elements. Since $C\subseteq A_5$, this proves that $C=A_5$. However, $G$ is clearly not solvable.

What a pity, the motivating idea for the question was attractive.

$\endgroup$
1
  • $\begingroup$ I've just recovered my previous account and now my new and old accounts are merged. I am sorry for this long delay of accepting your answer. I hope you will accept my excuse. $\endgroup$
    – Sh.M1972
    Feb 26, 2014 at 15:54
3
$\begingroup$

A remark in a special case.

Let $H$ be a finite group, $a$ a fixed point free automorphism of $H$, and $A = \langle a \rangle$. Let $G$ be the semidirect product of $H$ by $A$. Then it is well known that every element of $H$ is of the form $[a, h]$, for $h \in H$, so that $H = \gamma_2(G)$.

Now it is indeed true that $H$ (and thus $G$) is soluble in this case, but the proof requires the classification of finite simple groups.

$\endgroup$
1
  • $\begingroup$ @Andreas Caranti: yes and this point was my motivation in fact: it was proved by F. Ladish may be 4 years ago in Comm. Algebra using CFSG. I proved a version of it for Lie algebras using Cartan criterion (above mentioned paper). The above conjecture was one of my favorite problems during the past 3 years. $\endgroup$
    – Sh.M1972
    Dec 13, 2012 at 20:46

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.