Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

Let $p$ is a prime and $G$ is a finite $p$-group. Also, the normal subgroup $H$ of $G$ is maximal among abelian subgroups of $G$ which are normal in $G$ as well. Prove $H=C_G(H)$.

I should confess; I could do a few things about $G$ and $H$:

$Z(G)\subseteq H$,

$\frac{G}{C_G(H)}\leqslant S_{|H|}$.

So, please light my mind with your hints. Thanks.

share|cite|improve this question
You'll need to use that $G$ is solvable, since $C_2 \times 1 \leq C_2 \times A_5$ violates the conclusion. You'll need to use that $G$ is nilpotent, since $C_2 \leq \operatorname{SL}(2,3)$ also violates the conclusion. One nice feature of nilpotent groups is that their non-identity normal subgroups always contain non-identity central subgroups. Use this for $C_G(H)/H \leq G/H$ to get a contradiction if $C_G(H) > H$. – Jack Schmidt Aug 28 '12 at 14:00
@JackSchmidt: Sorry Jack for my terrible mistake. I forgot any $p$-group is a nilpotent one for a while, but do you assume $H$ is abelian? – Babak S. Aug 28 '12 at 15:19
Yes, I must assume $H$ is abelian for my argument to work. – Jack Schmidt Aug 28 '12 at 15:20
up vote 5 down vote accepted

Easier question

An easier question is, “If H is maximal amongst all abelian subgroups of G, then $C_G(H)=H$.”

Proof: Since H is abelian, $H \leq C_G(H)$, so we just need to rule out $H < C_G(H)$. Let $z \in C_G(H)$. Then $M = \langle H,z \rangle$ is abelian, since $z$ commutes with everything in $H$, and certainly commutes with itself. $H$ is central in $M$, and $M/H$ is cyclic (generated by $zH$), so $M$ is abelian. Thus $M=H$ by maximality and $z \in H$ and $C_G(H) \leq H$. Yay! $\square$

Your question

The problem with our proof for the question as asked is that $H$ is only assumed maximal amongst the normal abelian subgroups. Our “$M$” need not have been normal. We'd like to know that we can find some normal subgroup $M$ containing $H$ as a central subgroup with $M/H$ cyclic. How do we get such a small normal subgroup?

Non-identity normal subgroups of $p$-groups intersect the center nontrivially, so we know that in the group $G/H$, $C_G(H)/H$ must either be the identity, or it must contain a central element of order $p$, which will be our $zH$!


We begin by deciding which techniques are needed. The question could be asked for any group, but is only asked for $p$-groups. Why? Well, the conclusion is not true in general. The groups $\newcommand{\SL}{\operatorname{SL}}G=\SL(2,q)$ for $q \geq 3$ an odd prime power have only $H=Z(G)$ as a non-identity normal, abelian subgroup. Since $C_G(Z(G))=G$, clearly these groups violate the conclusion and so we must need to use some special property of $p$-groups. Since $\SL(2,3)$ is solvable, but not supersolvable we make the following conjecture (after checking for small counterexamples):

Proposition: If G is a supersolvable group and HG is maximal amongst abelian normal subgroups, then H is its own centralizer in G.

Indeed our proof is going to be the same: we just need to know that there is some $M/H$ with $M$ normal and $M/H$ cyclic. In a supersolvable group, the minimal normal subgroups are always cyclic.

Proof: Let $K=C_G(H)$ be the centralizer of H in G. Since H is normal in G, K is also normal in G. Since H is abelian, HK, and so we just need to rule out the possibility that H < K. Hence we consider $K/H \leq G/H$. Assume by way of contradiction that $K/H \neq 1$. Then $G/H$ has a minimal normal subgroup $M/H$ contained in $K/H$. Since $G$ is supersolvable and $M/H$ is a chief factor of $G$, $M/H$ is cyclic. Since $M \leq K$, $H$ is central in $M$, and so $M$ is an extension of its central subgroup $H$ by a cyclic group $M/H$. By the well known exercise, $M$ is abelian. Since $M$ is normal as well, this is a contradiction. $\square$

More generally, we just need G to act supersolvably on its Fitting subgroup. You can even allow the group to be infinite, though one needs to rephrase “minimal normal subgroup” slightly. I prefer the phrasing using minimal normal subgroups of quotient groups, also known as chief factors. Chief factors control a group “horizontally” by building the group in layers, and I highly recommend studying them.

Here a chief factor factor is a pair of normal subgroups $H < M$ of $G$ such that there is no normal subgroup $N$ with $H < N < M$; a chief factor is said to have a group theoretic property if the quotient $M/H$ has it, and it is said to have a subgroup theoretic property if $M/H$ has it in $G/H$. A finite group is supersolvable if and only if all of its chief factors are cyclic. A finite group is a $p$-group if and only if all of its chief factors are cyclic groups of order $p$. A finite group is solvable if and only if all of its chief factors are abelian. A group acts supersolvably on its normal subgroup $K$ if and only if every chief factor $H < M$ with $M \leq K$ is cyclic. The Fitting subgroup is the set of all elements that centralize every chief factor: $g \in G$ such that $[g,h] \in M$ for every $h \in H$.

share|cite|improve this answer

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.