Show that $H \leq C_G(C_G(H))$.

We have that $C_G(H) = \left\{ g \in G | gh = hg \; \forall h \in H \right\}$. So then $C_G(C_G(H))$ is the $g$'s that commute with all $g$'s that commute with $h$. This would mean $$C_G(C_G(H)) = \left\{ g \in G | gg'h = hg'g, \; \forall g' \in G, \; \forall h \in H \right\}$$ But since $gg' \in G$ this is just some element of $G$, so doesn't this become the same as $C_G(H)$?

I feel like what I have is wrong but I am not sure why.


Pick $h\in H$. We wish to show that $hg=gh$ for any $g\in C_G(H)$. But this is tautological from the definition of $C_G(H)$, so $H\subset C_G(C_G(H))$.

  • $\begingroup$ How does $hg=gh \implies H\subset C_G(C_G(H))$? $\endgroup$ – user372834 Oct 20 '18 at 17:49
  • $\begingroup$ @user372834 Just by writing down the definition of the centralizer $\endgroup$ – TomGrubb Oct 20 '18 at 17:52

in the definition $ C_G(C_G(H)) = \left\{ g \in G | gg'h = hg'g, \; \forall g' \in G, \; \forall h \in H \right\}$, is not $\forall g' \in G$, but $\forall g' \in C_G(H)$.

I believe that the question is wrong, because:

If $h_1\in C_G(C_G(H))$ and $h_1 \in H$, we have $$ h_1g'h_2=h_2g'h_1 ,\forall g' \in C_G(H), \forall h_2 \in H $$ It's clear that the identity $e \in C_G(H)$, so $$h_1h_2=h_2h_1, \forall h_2 \in H $$, then $H \subset C_G(C_G(H))$ only if H is abelian. But $H \leq C_G(C_G(H))$ also can indicate relation of order.

  • $\begingroup$ The definition of the centralizer of the centralizer was indeed written incorrectly. But the statement to be shown was correct. And I have never seen anyone use $\leq$ for order between subgroups. $\endgroup$ – Tobias Kildetoft Oct 21 '18 at 7:22
  • $\begingroup$ I got it! C_G(C_G(H)) = { g \in G | gg' = g'g, forall g' in C_G(H)} and it's easy to show . Thank you! The answer above is correct and I do not saw it. $\endgroup$ – Max Will Oct 21 '18 at 14:30

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.