This is an easy question for me to verify, I suppose, but, for whatever reason, I'm having a crisis of confidence.

This is a question.

The bit I'm unsure of is highlighted in bold text.

The Question:

Suppose $G$ is a group with $M,N\unlhd G$. Show that $MN\unlhd G$.

My Proof:

Since $M,N\unlhd G$, we have $e\in M\cap N$. Hence $e=ee\in MN$, so $MN\neq \emptyset$.

Let $g,h\in MN$. Then $g=mn$ and $h=m'n'$ for some $m,m'\in M$, $n,n'\in N$. Consider

$$\begin{align} gh^{-1}&=(mn)(m'n')^{-1}\\ &=(mn)(n'^{-1}m'^{-1}) \\ &=m(nn'^{-1})m'^{-1}. \end{align}$$

We may write $m=\mu m'$ for some $\mu\in M$ since $M$ is a group and right multiplication of an element of a group is a bijection on that group. Hence


but $m'(nn'^{-1})m'^{-1}\in N$ since $N\unlhd G$. Thus $gh^{-1}\in MN$.

Hence $MN\le G$ by the one-step subgroup test.

Now let $mn\in MN$ and $g\in G$. Then


where $g^{-1}mg\in M$ as $M\unlhd G$ and $g^{-1}ng\in N$ as $N\unlhd G$. Hence $g^{-1}mng\in MN$.

Thus $MN\unlhd G$. $\square$

Why am I unsure?

I don't know. What says that $\mu\in M$? I mean: I know it's because of the reason stated above but I feel like I've waved my hands a little bit there.

Please help :)

  • 2
    $\begingroup$ $\mu = m(m')^{-1}$ works $\endgroup$ – ZeroXLR Apr 18 at 14:24

Here is how you can formally justify the step you have a doubt on:

Consider the map

$$\phi_{m'}:M \to M: g\to gm'$$

This map is a bijection (an inverse is easy to construct, while constructing this you will see what $\mu$ actually is - also shown in the comments)

Thus, in particular, the map is surjective and thus for $m \in M$, there exists $\mu \in M$ such that $\mu m' =\phi_{m'}(\mu) = m$.

  • $\begingroup$ Ah, of course! Thank you :) $\endgroup$ – Shaun Apr 18 at 14:15
  • $\begingroup$ But does the highlighted bit make sense or have I sweeped stuff under the carpet? $\endgroup$ – Shaun Apr 18 at 14:16
  • $\begingroup$ I don't follow. The $m'$ is needed on the left hand side of $nn'^{-1}m'^{-1}$ in order for it to be a conjugate, so in $N$. $\endgroup$ – Shaun Apr 18 at 14:21
  • 1
    $\begingroup$ No, that's perfect. Thank you :) $\endgroup$ – Shaun Apr 18 at 14:26
  • 1
    $\begingroup$ No problem! Glad to help :) $\endgroup$ – EpsilonDelta Apr 18 at 14:27

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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