Is every coset of a group closed under taking inverses?

What I mean is that if $G$ is a group and $H$ is a subgroup, and let $a$ be any element of $G$. Then the coset $aH$ is not necessary a group. But does every element of $aH$ have an inverse element also in $aH$?

  • 3
    $\begingroup$ No, just take the coset $\{1\}$ of $\{0\}$ in $\mathbf{Z}/n\mathbf{Z}$ for $n\ge 3$. In general, for a group $G$ and subgroup $H$, one has (exercise): every left coset $gH$ is closed under inversion iff $H$ is normal and $G/H$ is an elementary $2$-group. $\endgroup$
    – YCor
    Sep 4, 2019 at 14:28

2 Answers 2


No, this does not have to hold.

Example: The symmetric group $S_3$ with 6 elements.

$U=\{\operatorname{id}, (12)\}$ is a subgroup.

Now view $(13)U=\{(13), (123)\}$. Then the element $(123)$ has no inverse.

We have $(123)(13)=(23)$ and $(123)(123)=(132)$


No. $a\in aH$. But if $a^{-1}\in aH$, then $a^{-1}=ah\implies h=a^{-2}$. So $a^2\in H$.

So, for instance, consider the dihedral group, $D_4=\langle r,s\mid r^4,s^2, (rs)^2\rangle $.

Take $H\le D_4$ where $H=\{s,e\}$. Then $r^2\not\in H$.

  • 1
    $\begingroup$ Indeed, $a^{-1} \in aH$ iff $a^2 \in H$. $\endgroup$
    – lhf
    Sep 4, 2019 at 12:07

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.