5
$\begingroup$

I am wondering whether there is a counterexample which shows that subgroups and quotients don't determine the group.

More precisely, suppose there are two groups $G_1, G_2$ such that all of their proper non-trivial normal subgroups are 1-1 corresponding and if $1<H_1 < G_1, 1<H_2 < G_2$ are those proper normal subgroups that they correspond, then $H_1 \simeq H_2$, and $G_1 / H_1 \simeq G_2/H_2$. (Here $\simeq$ means isomorphic.)

Then $G_1 \simeq G_2$?

I guess it might be not true in general, but I don't know any nontrivial counterexample except the pair $(\mathbb{Z}_p, \mathbb{Z}_q)$.

Any comments on this will be highly appreciated!

$\endgroup$
5
  • $\begingroup$ Not to be annoying, but if $G_1/H_1 \simeq G_2/H_2$ for any proper $H_1, H_2$, then you can take $H_1 = H_2 = 0$ and conclude, so you should exclude trivial subgroups from the hypothesis. $\endgroup$
    – qualcuno
    Aug 8, 2018 at 8:07
  • 1
    $\begingroup$ Since this seems to have been interpreted differently in the two answer, could you clarify if this is what you intended: Let $A$ and $B$ be the sets of proper non-trivial subgroups of the groups and assume we have a bijective map $f: A\to B$ such that we for all $H\in A$ have $H\cong f(H)$ and $G_1/H\cong G_2/f(H)$ whenever $H$ is normal? $\endgroup$
    – Tobias Kildetoft
    Aug 8, 2018 at 8:54
  • $\begingroup$ Following Tobias, what if the subgroups are not normal? Otherwise, why not simply take two simple groups? $\endgroup$
    – verret
    Aug 8, 2018 at 9:09
  • 1
    $\begingroup$ The question seems clear enough to me, but you need to specify that normal subgroups correspond to normal subgroup in the correspondence between subgroups, since otherwise you cannot form quotient groups. $\endgroup$
    – Derek Holt
    Aug 8, 2018 at 9:32
  • $\begingroup$ Thanks all. I corrected the original question as you commented. $\endgroup$
    – user29422
    Aug 8, 2018 at 14:04

3 Answers 3

Reset to default
10
$\begingroup$

There is a counterexample of order $605 = 11^2 \times 5$ with the structure $11^2:5$. Let $$G = \langle x,y,z \mid x^{11}=y^{11}=z^5=1, xy=yx, x^z=x^4, y^z=y^5 \rangle,$$ $$H = \langle x,y,z \mid x^{11}=y^{11}=z^5=1, xy=yx, x^z=x^4, y^z=y^3 \rangle.$$ These are $\mathtt{SmallGroup}(605,5)$ and $\mathtt{SmallGroup}(606,6)$ in the small groups database.

They both have $121$ conjugate subgroups of order $5$, two normal subgroups of order $11$ with nonabelian quotients of order $55$, $10$ non-normal subgroups of order $11$, $22$ non-normal subgroups of order $55$, and one normal subgroup of order $121$.

$\endgroup$
2
  • $\begingroup$ Aha Derek! That is the answer by the real master here. Excellent. +1 from me. $\endgroup$
    – Nicky Hekster
    Aug 8, 2018 at 10:09
  • $\begingroup$ Wow! This is the answer what I wanted. Thank you very much!:) $\endgroup$
    – user29422
    Aug 8, 2018 at 14:14
2
$\begingroup$

Smallest counterexample: $G_1=C_4$ generated by $x$ and $G_2=C_2 \times C_2$. Take $H_1=\langle x^2 \rangle$ and $H_2=C_2 \times \{1\}$.

$\endgroup$
4
  • 1
    $\begingroup$ The proper subgroups of these do not correspond one-to-one $\endgroup$
    – Tobias Kildetoft
    Aug 8, 2018 at 8:16
  • $\begingroup$ What does $C_n$ mean ?? $\endgroup$
    – Anik Bhowmick
    Aug 8, 2018 at 8:29
  • $\begingroup$ The cyclic group of order $n$ is denoted by $C_n$ (multiplicative notation) and this is also $\mathbb{Z}/n\mathbb{Z}$ (additive notation). But wait, Tobias triggered me: @user29422 what exaclty do you mean in your post by "such that all of their proper subgroups are 1-1 corresponding"? $\endgroup$
    – Nicky Hekster
    Aug 8, 2018 at 8:49
  • 1
    $\begingroup$ I added a comment on the question asking for a clarification from the OP. $\endgroup$
    – Tobias Kildetoft
    Aug 8, 2018 at 8:58
0
$\begingroup$

There are two (1) groups of order $4$. Each has a normal subgroup of order $2$ so the quotient groups are also order $2$. There is only one (1) group of order $2$ so the subgroups are isomorphic and the quotient groups are isomorphic.

(1) Up to isomorphism.

See The extension problem (Wikipedia)

$\endgroup$
4
  • $\begingroup$ The OP asked for the proper subgroups to correspond one-to-one $\endgroup$
    – Tobias Kildetoft
    Aug 8, 2018 at 8:16
  • 1
    $\begingroup$ @TobiasKildetoft Is there a technical meaning of "correspond" here that I am missing? $\endgroup$
    – badjohn
    Aug 8, 2018 at 8:21
  • $\begingroup$ Not sure what that would be. The two groups of order $4$ have different numbers of subgroups, so they don't correspond. $\endgroup$
    – Tobias Kildetoft
    Aug 8, 2018 at 8:22
  • $\begingroup$ @TobiasKildetoft I see what you mean now. So, it is a more interesting question than it first seemed. $\endgroup$
    – badjohn
    Aug 8, 2018 at 8:26

You must log in to answer this question.

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