It is well-known that there is an isomorphism between $PSp(4,3)$ (the symplectic group of dimension $4$ over $\mathbb F_3$) and $PSU(4,2^2)$ (the unitary group defined by $4\times4$ unitary matrices over $\mathbb F_4$).

Question: Can the exceptional isomorphism be interpreted in (finite) geometry?

Known examples of geometric interpretation:

The isomorphism $PSL(3,2)=PSL(2,7)$ is presented here.

The isomorphism $S_6=PSp(4,2)$ is explained in John Baez's article.

The isomorphism $A_6=PSL(2,9)$ can be explained by the Taylor graph of $PSL(2,9)$ acting on 10 points (i.e. the projective line of $\mathbb F_9$): this is a distance-regular graph on 20 vertices with intersection array $\{9,4,1;1,4,9\}$. Observing that this is also the intersection array of $J(6,3)$ and Johnson graphs are unique, it lets $S_6$ acts on the graph.

The isomorphism $PSU(3,3)=G_2(2)'$ can be explained by the unitary nonisotropics graph of $PΓU(3,q)$: for $q=3$, the intersection array is $\{6,4,4;1,1,3\}$, which is exactly that of a generalized hexagon of order 2. Furthermore, $GH(2,2)$ is unique up to duality, so it's also the Cayley generalized hexagon where $G_2(2)$ acts.

  • 3
    $\begingroup$ What's "finite geometry"?$$ $$ This group appears in three complex reflection groups; one is the Weyl group of $E_6$, whose reductions mod $2$ and $3$ identify it with $O_6^-(2)$ and $O_5(3)$; another is the complex (Eisenstein) $E_8$ lattice, whose reductions mod $2$ and $3$ take us to the same groups in their $U_4$ and $Sp_4$ guises which you asked about. Short vectors in each lattice give rise to several geometrical structures on which the group acts, including the $27$ lines of a cubic surface and the $36$ pairs $\pm r$ of $E_6$ roots. $\endgroup$ Dec 11, 2019 at 4:17
  • $\begingroup$ You forgot $A_8=GL(4,2)$ (which has a nice and simple geometric interpretation). $\endgroup$
    – abx
    Dec 11, 2019 at 5:35
  • 1
    $\begingroup$ @NoamD.Elkies this question seems your sort of thing, from yesterday: mathoverflow.net/questions/348159/… $\endgroup$
    – Will Jagy
    Dec 13, 2019 at 0:04

1 Answer 1


I believe that a finite geometric proof is given by Jean Dieudonné here:

Dieudonné, Jean, Les isomorphismes exceptionnels entre les groupes classiques finis, Can. J. Math. 6, 305-315 (1954). ZBL0055.01904.

If you go to Section 7 of the given paper you will see the proof. Dieudonné displays the required isomorphism using a 1-1 correspondence between the 40 lines in $(\mathbb{F}_3)^4$, and the non-isotropic lines in $(\mathbb{F}_4)^4$ (non-isotropic with respect to a non-degenerate hermitian form).

A google-books version is here.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

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