Does group isomorphism preserve characteristics subgroups?

122 Views Asked by At

So if $G_1$ and $G_2$ are isomorphic then $Z(G_1)$ and $Z(G_2)$ are isomorphic, more so the given an isomorphism $\phi: G_1 \to G_2$ then $\phi(Z(G_1)) = Z(G_2)$.

Is this true in general? That is for a general $\mathcal{S}$ that gives a characteristic subgroup $\mathcal{S}(G)$ of $G$ then is it true that any isomorphism $\phi$ then $\phi(\mathcal{S}(G_1)) = \mathcal{S}(G_2)$?

Is there hints on how to best tackle this problem? In the case of the center I use to property of the center of $G$, that is $\phi(xg) = \phi(gx)$ if $x \in Z(G)$ but don't even know how to begin for a general $\mathcal{S}$. Perhaps a counter example, I believe this property holds for well known characteristic subgroups like commutator subgroups (and Frattini subgroups as per comments).

1

There are 1 best solutions below

0
On BEST ANSWER

Intuitively, isomorphism preserves everything relevant. All an isomorphism does is relabel the things in a group. If you spraypaint your bike a different color, it is still functionally the same.

That is, if $S_1\leq G_1$ has some property relating to the group structure of $G_1$, then you should expect that $\phi(S_1)$ behaves the same way towards $G_2$. If your property is denotative or doesn't relate to the group structure, perhaps you shouldn't expect it to be preserved. Here are some examples:

If $G_1$ is cyclic then $G_2$ is cyclic.

If $G_1$ has a cyclic subgroup of order $n$ then $G_2$ has a cyclic subgroup of order $n$. Namely, if $\langle g_1\rangle \leq G_1$, then $\langle \phi(g_1)\rangle\leq G_2$ and it has the same order.

If $G_1$ is abelian then $G_2$ is abelian.

If $G_1$ has a nontrivial center then $G_2$ has a nontrivial center.

If $A$ is a subgroup of $G_1$ then $\phi(A)$ is a subgroup of $G_2$.

Etc.

This becomes clearer if you simply think of the definition of a homomorphism: $\phi(ab)=\phi(a)\phi(b)$. That is, however $a$ and $b$ interact in $G_1$, we know that $\phi(a)$ and $\phi(b)$ will interact in the same way in $G_2$. We are just relabelling $a$ to $\phi(a)$ and $b$ to $\phi(b)$. This relabelling is unique and covers all of $G_2$ since it is a bijection.

Sassatelli Giulio points out a counterexample to your (vague) general claim. The problem is indeed the claim's vagueness. Note that Sassatelli's counterexample is denotative; i.e. it relates to the way things are named rather than their structure. Of course, $\mathbb{Z}\cong C$, where $C$ is the (infinite) cyclic group generated by $c$. Technically, the underlying sets are different, but all relevant structural comparisons reveal that they are identical. An analogous property that refers to the group structure would be:

$\mathcal{S}(G) = \begin{cases} \{e_G\},\text{ if }G\not\cong\mathbb{Z}\\ \mathbb{Z},\text{ if }G\cong\mathbb{Z} \end{cases}$.

In this case, all groups isomorphic to $\mathbb{Z}$ are equal under $\mathcal{S}$. This is because equality is stricter than isomorphism. Equality demands not only that group structure is preserved, but also the names of the elements.

TL;DR just try to make sure that your property $\mathcal{S}$ relates to the group structure and it will probably be preserved through isomorphism.