What are the automorphisms of $SL_n$ as an algebraic variety?

In other words, let $k$ be an algebraically closed field of characteristic 0 (e.g., $k=\mathbb{C}$). Let $\tau$ be an automorphism of $SL_n$ regarded as an algebraic variety over $k$. Assume that $\tau$ takes the unit element $e$ of $G$ to itself. Is it true that $\tau$ is an automorphism of $SL_n$ as an algebraic group over $k$?

  • 1
    $\begingroup$ What about inversion (for $n>1$)? $\endgroup$ – ACL Jan 15 '13 at 16:27
  • $\begingroup$ @ACL: Thank you, Antoine. Are there any other automorphisms? $\endgroup$ – Mikhail Borovoi Jan 15 '13 at 16:33
  • $\begingroup$ @Mikhail: Your edit needs some more editing. Aside from this, what motivates the original question? $\endgroup$ – Jim Humphreys Jan 15 '13 at 17:04
  • $\begingroup$ @Jim: I have removed the edit. The original question was motivated by my previous question mathoverflow.net/questions/118356/… and a comment of Tom Goodwillie. I am trying to construct a finite subgroup $H\subset G=SL_{n,\mathbb{C}}$ and an automorphism $\sigma$ of $\mathbb{C}$ such that the $\mathbb{C}$-varieties $G/H$ and $\sigma(G/H)=G/\sigma H$ are not isomorphic. $\endgroup$ – Mikhail Borovoi Jan 15 '13 at 17:50
  • $\begingroup$ Note that the group generated by automorphisms, left translations and inversion is finite-dimensional (actually $2(n^2-1)$); while the example by Mariano gives a faithful action of an infinite dimensional abelian group. $\endgroup$ – YCor Jan 15 '13 at 18:10

The coordinate ring when $n=2$ is $A=k[a,b,c,d]/(ad-bc-1)$.

If $f\in k[b,c]$, there is an automorphism $\phi:A\to A$ such that $\phi(a)=a+bf$, $\phi(c)=c+df$, $\phi(b)=b$ and $\phi(d)=d$.

One could conjecture that the automorphism group in this case is generated by $SL_2$, inversion and this sort of triangular automorphisms, much as in the Makar-Limanov–Jung–van der Kulk theorem for $k[x,y]$ (This is a very optimistic conjecture, though: this is a $3$-dimensional affine variety quite close to affine space and there are non-tame automorphisms of the latter...)

In general, I doubt we know the automorphism group.

| cite | improve this answer | |
  • $\begingroup$ Of course, this trick works for all $n$. $\endgroup$ – Mariano Suárez-Álvarez Jan 15 '13 at 16:44
  • $\begingroup$ The polynomial f should be inside k[b, d]. $\endgroup$ – Anonymous May 16 '18 at 8:51
  • $\begingroup$ In fact, there are so-called "wild" automorphisms of k[a, b, c, d] / (ad-bc-1), i.e. automorphisms which do not belong to the subgroup generated by SL_2, inversion and the above mentioned "triangular" automorphisms. This is done in projecteuclid.org/euclid.jmsj/1359036456 $\endgroup$ – Anonymous May 16 '18 at 8:53

The automorphism group is massive!

Flexible varieties and automorphism groups, I. Arzhantsev, H. Flenner, S. Kaliman, F. Kutzschebauch, M. Zaidenberg, http://arxiv.org/abs/1011.5375.


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.