I am just learning to work with normal subgroups, and am trying to do the following proof. However, I have had some problems.

Let $N$ and $M$ be subgroups of a group $G$, where $M$ is a normal subgroup. Define $NM=\{nm : n \in N, m \in M\}$. We want to show that $NM$ is a subgroup of $G$.

In order to do this, we need to show three things: $NM$ is closed, it contains $G$'s identity element, and it has unique inverses.

Here is what I have:


Let $n_1m_1,n_2m_2\in NM$. I attempted to conclude from the fact that $M$ is normal that $(n_1m_1)(n_2m_2)=(n_1n_2)(m_1m_2)\in NM$, but have since learned that this does not work: Are all normal subgroups Abelian?

However, I am having trouble coming up with an alternate method. The fact that there exists an $m$ which commutes with the given $n$ doesn't seem to help prove the desired result in general, and likewise with any statement about the cosets $nM$ and $Mn$.


By the definition of a subgroup, $N$ and $M$ both contain $G$'s identity element, $e_G$. Thus, since $e_Ge_G=e_G$, we know that $e_G\in NM$.


I had a similar problem here as with closure. I wanted to say that $(nm)^{-1}=m^{-1}n^{-1}=n^{-1}m^{-1}\in NM$, the first step by the Socks-Shoes Property and the second step by the fact that $M$ is normal. This doesn't work for the same reason it doesn't work in proving closure, and I'm having trouble coming up with an alternate method for the same reasons.

  • 2
    $\begingroup$ You might also want to see The product of a subgroup and a normal subgroup is a subgroup. $\endgroup$ – Minus One-Twelfth Apr 22 at 0:50
  • $\begingroup$ Thank you very much for that link. Am I the only one who can't seem to find redundant questions on this site? I spent a good twenty minutes searching through results based on different relevant keywords before posting and didn't find that. In any case, thanks very much. $\endgroup$ – JustSomeGuy716 Apr 22 at 0:58

In general $(n_1m_1)(n_2m_2)\ne(n_1n_2)(m_1m_2)$. However, $M$ being normal in $G$ means that $gM=Mg$ for all $g\in G$. Hence $n_2M=Mn_2$. So we know that $n_2m_2\in n_2M=Mn_2$ and hence there is some element $m_3\in M$ such that $n_2m_2=m_3n_2$. So now $(n_1m_1)(n_2m_2)=n_1(m_2m_3)n_2$. Now again, $m_2m_3n_2\in Mn_2=n_2M$ and hence there is some $m_4\in M$ such that $m_2m_3n_2=n_2m_4$. So now $(n_1m_1)(n_2m_2)=n_1(m_2m_3)n_2=(n_1n_2)m_4\in NM$. That proves closure. Now you can do the tricks to prove that $NM$ contains inverses.


Since $M, N\le G$, we have $e=ee\in NM$. Hence $NM$ is nonempty.

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

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

We can write $n=\nu n'$ for some $\nu\in N$ since $N$ is a group and right multiplication by an element is a bijection on groups. Hence


but $n'(mm'^{-1})n'^{-1}\in M$ since $M\unlhd G$. Thus $gh^{-1}\in NM$.

Hence $NM\le G$ by the one-step subgroup lemma. $\square$


If N is a subgroup and M is a normal subgroup, then $NM = MN.$

This doesn't mean that every elements of $M$ commute with elements of M. (If it were true, we would say $M$ is a subset of the centralizer of $N$)

But it does mean that for any pair of elements $n_1,m_1,$ in $N,M$ respectively, there exists a pair of elements $n_2,m_2$ such that $m_2n_2 = n_1m_1.$ Possibly, but not necessarily, $n_1 = n_2$ or $m_1 = m_2$


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.