-2
$\begingroup$

If I have an (abelian) group $G$ and an automorphism $\sigma: G \to G$ then for any subgroup $H$ of $G$ there is an induced isomorphism $G/H \cong G/\sigma(H)$ given by the map $gH \mapsto \sigma(g)\sigma(H)$.

I want to know if the converse statement is true. Suppose I have an (abelian) group $G$ and two subgroups $H_1$ and $H_2$ such that $G/H_1 \cong G/H_2$ via some isomorphism. Then is there an automorphism $\sigma: G \to G$ which induces the automorphism on the quotient? I'm also interested in the slightly weaker question of if there is always an automorphism so that $\sigma(H_1) = H_2$ (ie an automorphism which induces some potentially different isomorphism on the quotients).

This feels plausible to me (I can't find a counterexample), but I haven't been able to construct the automorphism in general.

$\endgroup$
2
  • 1
    $\begingroup$ No, it is possible that $H_1$ is trivial but $H_2$ is not. Consider some infinite product. $\endgroup$
    – Wojowu
    Commented May 15, 2021 at 8:17
  • $\begingroup$ Did you check relatively easy cases, such as $G = {\bf Z}/p^2$ (where $p$ is a prime)? Does the number of automorphisms of $G$ exceed the number of proper subgroups? $\endgroup$ Commented May 15, 2021 at 14:15

1 Answer 1

2
$\begingroup$

The answer is rather obviously no for abelian groups. An infinite series of counterexamples can be found by taking $G=\mathbb{Z}_4 \times (\mathbb{Z}_2)^n,$ with $$H_1 = \mathbb{Z}_4 \times (\mathbb{Z}_2)^{n-1}, \quad H_2 = \mathbb{Z}_2 \times (\mathbb{Z}_2)^{n}.$$


Since you put "abelian" into parentheses, let me show that the answer is no also for non-abelian groups, providing a counterexample in which $H_1$ and $H_2$ are isomorphic as abstract groups. Consider the group of order $16$ $$G := \langle a,\,x \mid a^4 = x^4 = 1, \, xax^{-1} = a^{-1} \rangle,$$ namely, the non-trivial semidirect product of $\mathbb{Z}_4$ with itself, whose GAP label is $G(16, \, 4)$. Now, take the two subgroups $$H_1= \langle a^2, \, x \rangle, \quad H_2= \langle a, \, x^2 \rangle.$$ They are both isomorphic to $\mathbb{Z}_4 \times \mathbb{Z}_2$, so they are normal and the quotients are both isomorphic to $\mathbb{Z}_2$. However, $H_2$ is characteristic, so every automorphism of $G$ sends $H_2$ to itself.

$\endgroup$
2
  • 4
    $\begingroup$ There is also an abelian counterexample, $G=\mathbb Z_4\times\mathbb Z_2$. $\endgroup$
    – Wojowu
    Commented May 15, 2021 at 8:48
  • $\begingroup$ Actually, if the result were true, say for finite abelian groups, then it would follow that in an abelian group all subgroups of the same prime index are isomorphic, and by induction that all subgroups of the same cardinal are isomorphic. This is indeed false and $C_2\times C_4$ is the smallest counterexample (with non-isomorphic subgroups of order 4). $\endgroup$
    – YCor
    Commented May 15, 2021 at 10:06

Not the answer you're looking for? Browse other questions tagged .