Groups where there's always a "deformation retraction" homomorphism onto any subgroup.

291 Views Asked by At

Let $G$ be a finite group with the property that, for every subgroup $H$, there exists a homomorphism $f: G\to H$ such that $f(h)=h$ for all $h\in H$. What possible groups can $G$ be?

If $P$ is a Sylow-$p$ subgroup of $G$, then the homomorphism $f: G\to P$ has a kernel $K$. which is complementary to $P$ (the intersection is trivial since $f$ restricts to the identity on $P$, and $\langle P, K\rangle=G$, since $f(g)\in H$ for all $g\in G$, so $g^{-1}f(g)\in K$. Note that $K$ is normal. Now applying the assumption to $K$, we get a complement to $K$ with is normal itself and a Sylow-$p$, so the Sylow-$p$ is unique and normal, and so $G$ is the direct product of its Sylow-$p$ subgroups.

So now we just need to characterize the Sylow-$p$'s that satisfy this property, so now assume $G$ is a $p$-group. Let $H=Z(G)$. Then the deformation retraction homomorphism gives a normal complement to the center. But in a $p$-group, every nontrivial normal subgroup intersects the center nontrivially, so the complement to the center is trivial, so $G$ is abelian.

Now, if $G$ has a cyclic subgroup $C$ of order $p^2$, then let $H$ be the order $p$ subgroup of this cyclic subgroup. Then a generator $x$ of $C$ can't map to the identity (otherwise all of $H$ would map to the identity), so it must map to a generator $y$ of $H$. But then $x^p$ would have to map to $y^p=1$, which is a contradiction, since $x^p\in H-\{1\}$ (it's in $H$ since it's the $p$th power of something in $C$, and it's not the identity, since $x\notin H$). Thus, $G$ cannot have a cyclic subgroup of order $p^2$, and thus must be elementary abelian. If so, then $G$ does in fact have this property, since it's just a vector space over $\mathbb F_p$, and you can just project onto any subspace.

Therefore, the only $G$ with the deformation detraction property are those that are the direct products of prime cyclic groups. However, I was wondering if there were an elementary way to do this, since apparently, this problem was from before Sylow subgroups or the Fundamental Theorem of Finite Abelian Groups were covered.

1

There are 1 best solutions below

0
On BEST ANSWER

For any subgroup $K$ let $f_K\colon G\to K$ be a homomorphism that restricts to the identity on $K$, and let $i_K\colon K\to G$ be the inclusion. Then we have a short exact sequence $$0\longrightarrow \ker f_K \overset{i_{\ker f_K}}{\longrightarrow} G\overset{f_K}{\longrightarrow} K\longrightarrow 0.$$ This sequence splits by using $f_{\ker K}$ and $i_K$. Therefore $G=K\times \ker f_K$ for all subgroups $K$. Specifically, every subgroup is a direct factor, from which it follows that every irreducible factor is a cyclic group of prime order and $G$ is a direct product of such groups.

Note the only part that requires finiteness is that $G$ can be written as a product of irreducibles. Finiteness can then be replaced by chain conditions by Krull-Schmidt.