MathOverflow is a question and answer site for professional mathematicians. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

Is $PSL(n, \mathbb Z)$ isomorphic to a subgroup of $GL(n,\mathbb C)$ or even $GL(n+1,\mathbb C)$?

share|cite|improve this question
When $n$ is odd, $PSL(n,\mathbb Z)$ is the same as $SL(n,\mathbb Z)$. So yes to the first question in that case. – Tom Goodwillie Apr 23 '11 at 23:04
It's an intriguing question all by itself, but it might be equally interesting to know what kind of consequences the answer would have. – Jim Humphreys Apr 25 '11 at 15:29

You can look for a finite subgroup of $PSL(n,\mathbb Z)$ such that every faithful representation of it has dimension $>n+1$. The following works for even $n\ge 6$.

In $GL(n,\mathbb Z)$ take the group of monomial matrices, i.e. products of permutation matrices and diagonal matrices. Intersect with $SL(n,\mathbb Z)$ and let $G$ be the image in $PSL(n,\mathbb Z)$.

$G$ maps onto the symmetric group $S_n$ with kernel $D\cong (\mathbb Z/2)^{n-2}$. In any faithful action of $G$ on $\mathbb C^d$ there is a one-dimensional subspace $L$ acted on nontrivially by $D$. In the action of $S_n$ on the set of nontrivial $1$-dimensional characters of $D$ every orbit has at least $n(n-1)/2$ elements. It follows that there at least that many independent choices of $L$ and therefore $d\ge n(n-1)/2>n$.

One step breaks down when $n=4$, so you'd need a different subgroup in that case.

share|cite|improve this answer
Thanks Tom. I was, of course, only interested in the n even case. Your answer definitively answers the question for my purposes. – John Franks Apr 24 '11 at 15:39
Rather than add this as an answer it would be preferable just to accept Tom's answer. – Jim Humphreys Apr 25 '11 at 15:21
This is definitely a creative approach, though I still wonder whether something more conceptual (in a more general context of arithmetic groups) is possible. Also, I wonder what an efficient subgroup choice would be for $n=4$? – Jim Humphreys Apr 25 '11 at 15:25
I think you can use a subgroup of $PSL(4,\mathbb Z)$ that is conjugate in $PSL(4,\mathbb R)$ to a subgroup of $PSO(4)\cong SO(3)\times SO(3)$ that is isomorphic to $A_4\times A_4$. – Tom Goodwillie Apr 26 '11 at 1:25

If $n \ge 3$, perhaps one can use Margulis superrigidity to deduce that any homomorphism from $\mathrm{SL}(n,\mathbb{Z})$ to $\mathrm{GL}(m,\mathbb{C})$ extends to a representation of $\mathrm{SL}(n,\mathbb{C})$. The standard representation of $\mathrm{SL}(n,\mathbb{Z})$ factors through $\mathrm{PSL}(n,\mathbb{Z})$ if and only if $n$ is odd. The next smallest representation of $\mathrm{SL}(n,\mathbb{C})$ has dimension $> n+1$, (assuming that $n > 2$). If $n = 2$, then $\mathrm{PSL}(2,\mathbb{Z})$ is certainly a subgroup of $\mathrm{GL}(3,\mathbb{Z})$. So the answer seems to be: if and only if $n = 2$ or $n$ is odd.

share|cite|improve this answer
What Margulis superrigidity tells you is that a homomorphism $SL_n(\mathbb{Z})\rightarrow GL_m(\mathbb{C})$ is a direct sum of a homomorphism extending to $SL_n(\mathbb{R})$ and of a homomorphism factoring through some congruence subgroup. – Alain Valette Apr 24 '11 at 9:05

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.