2

Since I still don’t know the answer, I’ve also asked it on math.overflow.

I saw a remark in an old post that

$G/F(G)$ is isomorphic to a group of the form $X_1 \times \ldots \times X_t,$ where each $X_i$ is isomorphic to a subgroup of a completely reducible solvable subgroup of ${\rm GL}(n_i,p_i)$ for some integer $n_i$ and prime $p_i$.

Before I ask my question, I need to fill in some details.

It is well-known that for a finite solvable group $G$, $G/F(G)$ is isomorphic to a subgroup of ${\rm Out}(F(G))$. Let $p_i(i=1,\cdots, t)$ be all the prime divisors of $|G|$. It follows that $G/F(G)$ is isomorphic to a subgroup of ${\rm Out}(O_{p_1}(G))\times\cdots\times{\rm Out}(O_{p_t}(G))$.

If $\Phi(G)=1$, then $\Phi(O_{p_i}(G))\le\Phi(G)=1$ for each $p_i$ dividing $|G|$. It follows that each $O_{p_i}(G)$ is elementary abelian, namely there exists a positive integer $n_i$ such that $O_{p_i}(G)=C_{p_i}^{n_i}$. Since we can see elementary abelian groups as vector spaces, ${\rm Aut}(O_{p_i}(G))\cong{\rm GL}_{n_i}(\mathbb{F}_{p_i})$. In addition, since $O_{p_i}(G)$ is abelian, ${\rm Inn}(O_{p_i}(G))$ is trivial and hence ${\rm Out}(O_{p_i}(G))\cong {\rm Aut}(O_{p_i}(G))\cong{\rm GL}_{n_i}(\mathbb{F}_{p_i})$. Then we know that $G/F(G)$ is isomorphic to a subgroup of ${\rm GL}_{n_1}(\mathbb{F}_{p_1})\times\cdots\times{\rm GL}_{n_t}(\mathbb{F}_{p_t})$.

If $\Phi(G)\neq 1$, we still can know that $\Phi(G/\Phi(G))$ is trivial. Since $G$ is solvable, we have $\Phi(G)\subseteq F(G)$ and $F(G)/\Phi(G)=F(G/\Phi(G))$. By the third isomorphism theorem, $$G/F(G)\cong G/\Phi(G)\big/F(G)/\Phi(G)= G/F(G)\cong G/\Phi(G)\big/F(G/\Phi(G)) .$$Hence we can get a similar conclusion.

My question is: By the reasoning above, I can only know that $G/F(G)$ is isomorphic to a subgroup of ${\rm GL}_{n_1}(\mathbb{F}_{p_1})\times\cdots\times{\rm GL}_{n_t}(\mathbb{F}_{p_t})$. In general, if $A$ is a subgroup of $A_1\times\cdots\times A_t$, we cannot say $A$ is of the form $X_1\times\cdots\times X_t$, where $X_i$ is isomorphic to a subgroup of $A_i$. But here each $X_i$ is isomorphic to a subgroup of ${\rm GL}(n_i,p_i)$. I want to know what was different in this specific case.

Thanks!

  • 1
    You seem to be assuming that the group $G$ is solvable – Derek Holt May 20 '20 at 07:28
  • @DerekHolt Yes. In my linked posts they were all assumed to be solvable. I’ve edited my question to clarify that. Did I miss anything helpful for my question when $G$ is solvable? –  May 20 '20 at 07:30
  • As stated, the assertion is trivially true for every finite solvable group, because we can just choose a single prime $p_1=p$ that does not divide $|G|$ (so every subgroup of ${\rm GL}(n,p)$ is completely reducible). Perhaps the $p_i$ are intended to be the primes dividing $|F(G)|$? – Derek Holt May 20 '20 at 07:41
  • @DerekHolt Thanks for your comment. I’m not familiar with the term “completely reducible”. Could you please tell me how it helps me with this question? –  May 20 '20 at 07:45
  • @DerekHolt Yes, $p_i$ are intended to be the primes dividing $|F(G)|$, because $p_i$ are all prime divisors of $|G|$, and $|G|$ and $|F(G)|$ have exactly the same set of prime divisors. –  May 20 '20 at 07:51
  • 2
    You should edit the question to include all of the assumptions. – Derek Holt May 20 '20 at 08:28
  • @DerekHolt I’ve just edited my question and I think all the assumptions are included now. –  May 20 '20 at 08:38
  • I don't see any edits. The quoted remark at the top of your post is only true if $G$ is solvable, and in that case, as I explained before, it is trivially true. – Derek Holt May 20 '20 at 10:38
  • Now posted to MO, https://mathoverflow.net/questions/360865/g-fg-is-isomorphic-to-x-1-times-cdots-times-x-t – Gerry Myerson May 20 '20 at 12:59
  • The original answer was slightly misstated: It should have said that $G/F(G)$ is A SUBGROUP OF a group of the form $X_1 \times X_2 \times \ldots X_t$, where each $X_i$ is a completely reducible subgroup of …... – Geoff Robinson May 20 '20 at 16:41

1 Answers1

4

As I have said in comments, it is not hard to prove the remark as stated.

But here is a stronger statement that is definitely not true.

Let $G$ be solvable and suppose that $F(G)$ is a direct product of elementary abelian groups $C_{p_i}^{n_i}$ for fprimes $p_1,\ldots,p_t$. Then $G/F(G)$ is isomorphic to a direct product $X_1 \times \cdots \times X_t$, where $X_i$ is isomorphic to a solvable completely reducible subgroup of ${\rm GL}(n_i,p_i)$.

That is not true. I can construct an example in which $F(G) = C_2^2 \times C_5^2$. and $G/F(G)$ is isomorphic to a subgroup of index $2$ in$X_1 \times X_2$ with $X_1 \cong X_2 \cong S_3$, but $G/F(G)$ is indecomposable and is not itself isomorphic to $X_1 \times X_2$ for any $X_1 \le {\rm GL}(2,2)$ and $X_2 \le {\rm GL}(2,5)$.

Derek Holt
  • 90,008
  • Yes, there was a slip in the original answer: It should have said that $G/F(G)$ is isomorphic to A SUBGROUP OF a group of the form $X_1 \times X_2 \times \ldots \times X_t $, where each $X_i$ is a completely reducible solvable subgroup of ${\rm GL}(n_i,p_i)$ – Geoff Robinson May 20 '20 at 16:50