I am struggling with this question: show that there does not exist functor from $Group$ to $Set$ taking each group to its set of automorphisms. I have thought about it for a while now, not having any insight.

There is a thread on MSE showing the nonexistence of a functor taking each $\mathcal{G}$ in $Group$ to Aut($\mathcal{G}$) in $Group$, which is relevant but not of help.

Edit: I know that for some 1-1 homomorphism $f:G\longrightarrow H$, there is an automorphism $A$ on $G$ such that there is more than one automorphisms $g_1$ and $g_2$ on $H$, for which $g_1\circ f=g_2\circ f=f\circ A$.

On the other hand, for some 1-1 homomorphism $f:G\longrightarrow H$, there is an automorphism $A$ on $G$, such that there is no automorphism $g$ on $H$ for which $g\circ f=f\circ A$

This rules out one way of defining the functor, but I am not sure how it helps in the general case.


I have thought of one way of doing this, but there might be easier ways. Let $\mathcal{F}$ be a functor with the properties you describe.

Suppose that we have groups $G$, $H$ with homomorphisms $\phi:G \to H$ and $\psi:H \to G$ with $\psi \phi = {\rm Id}_G$. Then $\mathcal{F}(\psi\phi) = \mathcal{F}(\psi) \mathcal{F}(\phi)$ is the identity map on $\mathcal{F}(G)$. So if we can find such an example with $|{\rm Aut}(H)| < |{\rm Aut(G)}|$, then this is not possible, and we have a contradiction. Note that, in this situation, $H = K \rtimes G$ with $K = \ker(\psi)$.

One such example is $G=C_2^n$ elementary abelian of order $2^n$ and $H = S_3^n$. Then ${\rm Aut}(G) = {\rm GL}(n,2)$ and ${\rm Aut}(H) = S_3 \wr S_n$. We have $|{\rm Aut}(H)| < |{\rm Aut(G)}|$ for $n \ge 5$.

  • $\begingroup$ Thanks for the answer. Could you please explain what $H=S^3_n$ is? $\endgroup$ – Chase Mar 8 '14 at 17:09
  • 2
    $\begingroup$ It is the direct product of $n$ copies of the symmetric group $S_3$ (or the dihedral group $D_6$ if you prefer). It's not too hard to show that its automorphism group must permute these $n$ direct factors, so we get $|{\rm Aut}(H)| = 6^nn!$. $\endgroup$ – Derek Holt Mar 8 '14 at 17:59

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.