0

Let $K$ is a normal subgroup of a group $G$,

if $G \cong G/K$, $K=\{ e \}$?

Firstly, the converse is true.

If $G$ is a finite group, it is obviously true.

But if $G$ is an infinite group, I've no idea how to prove it or to come up with any counterexample.

Bernard
  • 175,478

2 Answers2

2

Counterexample: Say $G$ is the group of sequences of integers, with addition. Let $K$ be the subgroup consisting of sequences $(n_1,0,0,\dots)$.

In slightly more detail: An element of $G/K$ is a coset $K+(n_1,n_2,\dots)$. Define $\phi:G\to G/K$ by $\phi((n_1,n_2,\dots))=K+(0,n_1,n_2,\dots)$. Then you can show $\phi$ is an isomorphism. (It's clear that $\phi(n+m)=\phi(n)+\phi(m)$ and that $\phi$ is surjective; it's not hard to see that if $\phi(n)=0$, that is, $\phi(n)=K+(0,0,\dots)$, then $n=0$; hence $\phi$ is injective.)

Oops, maybe it's not clear that $\phi$ is surjective. Regarding that, note that $K+(n_0,n_1,n_2,\dots) = K+(0,n_1,n_2,\dots)$.

1

No, take $\mathbb{Z}^{\times \mathbb{N}}$ and the subgroup $(n,0,...0)$