15
$\begingroup$

I define a 1-isomorphism between two groups as a bijection that restricts to an isomorphism on every cyclic subgroup on either side. There are plenty of examples of 1-isomorphisms that are not isomorphisms. For instance, the exponential map from the additive group of strictly upper triangular matrices to the multiplicative group of unipotent upper triangular matrices is a 1-isomorphism. Many generalizations of this, such as the Baer and Lazard correspondences, also involve 1-isomorphisms between a group and the additive group of a Lie algebra/Lie ring.

Consider the following function F associated to a finite group G. For divisors $d_1$, $d_2$ of G, define $F_G(d_1,d_2)$ as the number of elements of G that have order equal to $d_1$ and that can be expressed in the form $x^{d_2}$ for some $x \in G$.

Question: Suppose G and H are finite groups of the same order such that $F_G = F_H$. Does there necessarily exist a 1-isomorphism between G and H?

Note that the converse is obviously true: if there exists a 1-isomorphism between G and H, then $F_G = F_H$.

Incidentally, just knowing the orders of elements does not determine the group up to 1-isomorphism. There are many counterexamples of order 16, with two non-abelian groups (one being the direct product of the quaternion group and the cyclic group of order two, and the other a semidirect product of cyclic groups of order four) having the same statistics on orders of elements as $\mathbb{Z}_4 \times \mathbb{Z}_4$, but neither being 1-isomorphic to it because they don't have the same number of squares.

Similarly, just knowing how many elements are there of the form $x^d$ for each divisor d of the order is not sufficient to determine the group up to 1-isomorphism. Again, there are counterexamples of order 16.

$\endgroup$
5
  • $\begingroup$ What is the easiest example of non-isomorphic finite groups $F$ and $G$ such that $F_g = F_h$? $\endgroup$ Jul 23, 2010 at 0:17
  • $\begingroup$ Any examples arising from the Lazard or Baer correspondences or their generalizations would work. For instance, for odd primes $p$, there is a bijection between the multiplicative group $U(3,p)$ of upper triangular unipotent $3 \times 3$ matrices over the field of $p$ elements, and the elementary abelian group of order $p^3$. There are also some examples of order $2^4$, which don't strictly arise from the Lazard correspondence but arise from generalizations of it. $\endgroup$
    – Vipul Naik
    Jul 24, 2010 at 19:22
  • 2
    $\begingroup$ Is a 1-isomorphism any different from a bijection f satisfying $f(x^n)=f(x)^n$ for integers n? $\endgroup$ Jul 24, 2010 at 22:29
  • $\begingroup$ They look to be the same, yes. $\endgroup$ Jul 25, 2010 at 0:53
  • $\begingroup$ George Lowther: Your definition of 1-isomorphism matches what I wrote. $\endgroup$
    – Vipul Naik
    Jul 27, 2010 at 13:52

2 Answers 2

16
$\begingroup$

Here is a counterexample of order $32$.

$G$ and $H$ will each have $3$ elements of order $2$ and $28$ elements of order $4$. In both cases all three elements of order $2$ will have square roots. That insures that $F_G=F_H$. But in $G$ one of them will have $4$ square roots while the others each have $12$, and in $H$ one of them will have $20$ square roots while the others each have $4$. That rules out a $1$-isomorphism.

Let $Q$ be a quaternion group of order $8$ and let $C\subset Q$ be a (cyclic) subgroup of order $4$. Inside $Q\times Q$ there are three subgroups of index $2$ that contain $C\times C$. Let $G$ be $Q\times C$ and let $H$ be the one that is neither $Q\times C$ nor $C\times Q$.

$\endgroup$
5
  • $\begingroup$ it always seems to come back to the difference between the dihedral, quaternion and semi-dihedral groups! $\endgroup$ Jul 28, 2010 at 3:36
  • $\begingroup$ I don't think of this example as being about the difference between quaternion and dihedral. I didn't think about counterexamples of order $p^5$ for other primes. $\endgroup$ Jul 28, 2010 at 12:43
  • $\begingroup$ Tom Goodwillie: Thank you for the answer. I'm sorry I was away from Math Overflow for the last few days, so did not notice this answer from you. Your answer seems to have been auto-selected as the corrected answer. $\endgroup$
    – Vipul Naik
    Aug 5, 2010 at 20:34
  • $\begingroup$ I didn't know that auto-selection was possible. $\endgroup$ Aug 6, 2010 at 1:09
  • $\begingroup$ If a question has a bounty, the highest-voted answer will be auto-accepted when the bounty period runs out if none has been accepted by then. $\endgroup$ Aug 7, 2010 at 1:53
2
$\begingroup$

Not a complete answer, but a way of breaking the question down a bit.

We put a relation $R$ on a finite group $G$ by saying $x <_R y$ if $x$ is contained in the subgroup generated by $y$. This relation is preserved by 1-isomorphisms. Moreover, we can recover $F_G$ from $R$: the order of an element $x$ is the number of elements $y$ for which $y <_R x$, and $x$ is a $d$-th power if and only if there is a $y$ such that $x <_R y$ and $|y|=(d,|y|)|x|$.

  1. Can we determine the $R$-structure from $F_G$? In other words, if $F_G = F_H$, does this ensure there is an $R$-preserving bijection from $G$ to $H$?

  2. Suppose there is a bijection from $G$ to $H$ compatible with $R$. Are $G$ and $H$ 1-isomorphic? (It's easy to see that $R$-preserving maps are not always 1-isomorphisms: for instance $C_5$ has $24$ $R$-preserving permutations, but only $4$ (1-)automorphisms.)

There has also been some work on power graphs of finite groups which may be relevant here: the power graph has vertex set $G$ and an edge from $x$ to $y$ if $x <_R y$ or $y <_R x$. It was proved recently by Peter Cameron that this symmetric relation is enough to recover the $R$-structure. See: http://www.reference-global.com/doi/abs/10.1515/JGT.2010.023

$\endgroup$
2
  • $\begingroup$ I think the answer to (2) is yes. Basically, we can use the relation R to determine a new symmetric relation where $x \sim y$ if and only if $x <_R y$ and $y <_R x$. This new symmetric relation just means that the two elements generate the same cyclic subgroup. This equivalent relation divides the group into orbits. We can then construct a directed acyclic graph on the set of these orbits and then get a bijection between the DAGs for the two groups, which then can be extended to a 1-isomorphism. So, the power graph seems to be a combinatorial object that stores the group up to 1-isomorphism. $\endgroup$
    – Vipul Naik
    Aug 6, 2010 at 16:13
  • $\begingroup$ Yes: I suppose this is saying that although an R-isomorphism isn't necessarily a 1-isomorphism, we can make it into one by permuting elements within the equivalence classes. In the other direction, a 1-isomorphism from $G$ to some known group is certainly enough to determine the power graph of $G$ (not just up to isomorphism). The next natural question is: which finite graphs are power graphs of finite groups? This could well be an active area of research, but I don't know how far people have got with it. $\endgroup$
    – Colin Reid
    Aug 9, 2010 at 10:29

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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