Let $|G|=p^n$ and $p$ a prime and let $|G:C_G(x)|\leq p$ for all $x\in G$. Prove:

$(a)~~~~~C_G(x)\trianglelefteq G$ for all $x\in G;$

$(b)~~~~~G’\leq Z(G);$

${\color{red}{{(c)~~~~~|G’|\leq p}.}}$

Edit: I’m now only confused about $\bf (c)$. For $\bf(c)$, the nice answers below are not complete. I was told that $\bf (c)$ has something to do with this paper (or here Knoche,H.G.: Überden Frobeniusschen Klassenbegriff in nilpotenten Gruppen, Math. Z. 55 (1951), 71–83. ), but reading the whole paper would be difficult for me, since I’m no German speaker; more unfortunately, I have not even been able to figure out which part I need to read. How does that paper help? I’m quite a beginner in group-theory, and I would be grateful if I could receive your instruction.

  • 2
    $\begingroup$ @MCT If $x\in Z(G)$ then $C_G(x)=G$, hence $C_G(x)\trianglelefteq G$. Anything wrong? $\endgroup$ – Benny Mar 1 '18 at 2:16
  • 1
    $\begingroup$ You may want to check this entry math.stackexchange.com/questions/816442/… $\endgroup$ – Nicky Hekster Mar 1 '18 at 7:52
  • $\begingroup$ @NickyHekster : ) I’ve found that link myself. Haha, thanks all the same! $\endgroup$ – Benny Mar 1 '18 at 7:55
  • $\begingroup$ @NickyHekster By the way, how does that link help me with $(c)$? $\endgroup$ – Benny Mar 1 '18 at 8:01
  • $\begingroup$ OK, let me give it a try. $\endgroup$ – Nicky Hekster Mar 1 '18 at 8:09

For $(c)$ you need Satz 2 at the bottom of page 84; this applies in case the index of the normalizer of any element of $G$ is at most $p$ (which the paper phrases as $r=1$).

The fact that $r\leq1$ follows immediately from the premise that $|G:C_G(x)|\leq p$ for all $x\in G$. The separate case that $r=0$ is easy as then $G$ is abelian.

As in the paper, let $r$ be the smallest integer for which $|G:N_G(x)|\leq p^r$ holds for all $x\in G$, where $N_G(x)$ denotes the normalizer of $x$ in $G$. Because $C_G(x)\subset N_G(x)$ and $|G:C_G(x)|\leq p$ for all $x\in G$ it follows that $r\leq1$. Here's a rough paraphrase of Satz 2 and the relevant half of the proof:

Theorem 2. Let $G$ be a finite $p$-group. Then $r=1$ if and only if $|G'|=p$.

Proof. Let $K:=[A,B]\neq e$ be a commutator of two elements of $G$, and let $[A',B']$ be another commutator distinct from $e$ with $A',B'\in G$. Then we choose another element $A''\in G$ that is not contained in $N_G(B)$ and $N_G(B')$. Under these assumptions $A''$ lies in $A^iN_G(B)$ for some $i\neq0\pmod{p}$, and so because $G'\subset Z(G)$ we get $$[A'',B]=[A,B]^i=K^i.$$ Similarly there exists $j\neq0\pmod{p}$ such that $$[A'',B']=[A'',B]^j=K^{ij},$$ and finally $$[A',B']=[A'',B']^k=K^{ijk}.$$ We see that the commutator of an arbitrary pair of group elements can be written as a power of one and the same element $K$. As every $p$-th power lies in $Z(G)$, also the $p$-th power of a commutator equals $e$. Hence $G'$ is cyclic of order $p$.

[This proves one half of the theorem, the other half I have omitted.]

In the remaining case that $r=0$, clearly $G$ is abelian and hence $|G'|=1$.

  • $\begingroup$ Thanks! But what does Staz 2 say? $\endgroup$ – Benny Mar 10 '18 at 15:42
  • 1
    $\begingroup$ Added a rough translation; not sure if it is entirely self-contained. Let me know if something is missing. $\endgroup$ – Servaes Mar 10 '18 at 15:55
  • $\begingroup$ Why every $p$-th power lies in $Z(G)$? $\endgroup$ – Benny Mar 10 '18 at 16:20
  • $\begingroup$ Because $r=1$, for every $x\in G$ every $p$-th power is in $N_G(x)$. Hence every $p$-th power is in the intersection of all these normalizers, which is $Z(G)$. $\endgroup$ – Servaes Mar 10 '18 at 16:29

For $(a)$, what you need to show is something more general: any subgroup of index $p$ in a $p$-group $G$ is a normal subgroup of $G$.

For $(b)$ remark that $G/C_G(x)$ is abelian for all $x$ thus

$G'\leq C_G(x)$ (by definition $G'$ is contained in every normal subgroup $N$ of $G$ such that $G/N$ is abelian).

And then use the equality :

$\bigcap_{x\in G}C_G(x)=Z(G)$.

For $(c)$, the Frattini subgroup is defined as the intersection of all maximal subgroups. Maybe you should stay simple. Although I quite don't see why, I would be inclined to think that you want to show that either $G=Z(G)$ or $|Z(G)|=p$.


For part (a), you can use this corollary.

Corollary. If $G$ is a finite group of order $n$ and $p$ is the smallest prime dividing $|G|$, then any subgroup of index $p$ is normal.

Here if $|G:C_{G}(x)|=p$, then by the corollary, $C_{G}(x)\unlhd G$.


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.