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


This question somehow is related to a previous question I asked here. Let me set some notations. Let $\mathbb{F}_p$ be a finite field with $p$ element. Consider the following symmetric matrix

\begin{equation} J:=\begin{pmatrix} 0 & I_n \\\\ I_n & 0 \end{pmatrix}, \end{equation}

where $I_n$ is the identity matrix. Now the special orthogonal group is defined by

$$SO_{2n}(\mathbb{F}_p):=\{A\in SL_n(\mathbb{F}_p): AJA^T=J \}.$$

Obviously, when $\sigma,\tau\in M_n(\mathbb{F}_p)$ are skew-symmetric the following matrices belongs to the special orthogonal group. \begin{equation} \begin{pmatrix} I_n & \sigma\\\\ 0 & I_n \end{pmatrix}, \begin{pmatrix} I_n & 0\\\\ \tau & I_n \end{pmatrix}\in SO_{2n}(\mathbb{F}_p) \end{equation}

I have done some computations to show that these matrices generate $SO_{2n}(\mathbb{F}_p)$. But now, based on the question I have asked, I might have done some mistakes in my computations. Is it true that matrices mentioned above are a generating set?

share|cite|improve this question
In even characteristic the set of invertible $A$ s.t. $AJA^\top=J$ would give you the symplectic group rather than an orthogonal group. – Dima Pasechnik Jul 15 '12 at 12:12

When $n=1$ then your matrices $\sigma$ and $\tau$ must be zero (since they are skew-symmetric), and hence your two generators are equal to one. But $-id\in SO_{2n}(\mathbb F_p)$, so the group is not actually trivial.

But even if $n>1$ there is nothing that keeps you from choosing $\sigma=\tau=0$. So maybe you want to at least consider all matrices of the given form.

Edit: So, following the comments, I now assume that you let $\sigma$ and $\tau$ range over all skew-symmetric matrices (instead of just picking two; however it still suffices to let them range over a basis). Still, for $n=2$, the group generated by the matrices from the question is isomorphic to $SL_2(\mathbb F_p)$. Now one just has to compare orders to see that this isn't $SO_{4}(\mathbb F_p)$. Or just use GAP:

gap> A := One(GF(5))*[[1,0,0,0],[0,1,0,0],[0,-1,1,0],[1,0,0,1]];;
gap> B := One(GF(5))*[[1,0,0,1],[0,1,-1,0],[0,0,1,0],[0,0,0,1]];;
gap> G := Group(A, B);
<matrix group with 2 generators>
gap> Size(G);
gap> Size(SO(1,4,5)); 

Edit 2: The matrices don't necessarily generate the group for $n=3$ either (checked for $p=5$ and $p=7$; for bigger $n$ the computation take longer than a few seconds so I didn't go there):

gap> n := 3;; p := 5;;
gap> f := sigma -> AsList(BlockMatrix([[1,1,IdentityMat(n)], [1,2,sigma],[2,1,NullMat(n,n)], [2,2,IdentityMat(n)]], 2, 2));
function( sigma ) ... end
gap> M := Concatenation(List([1..n], i -> List([i+1..n], j -> List([1..n], k -> List([1..n], function(l) if k = i and j = l then return 1; elif k = j and l = i then return -1; fi; return 0; end)))));;
gap> G := Group(One(GF(p))*Concatenation(List(M, f), List(M, v -> TransposedMat(f(v)))));
<matrix group with 6 generators>
gap> Size(G); Size(SO(1,2*n,p)); Size(SO(1,2*n,p)) mod Size(G);
share|cite|improve this answer
I meant for $n>1$. Would you please give more details by your second paragraph? I am not sure if I understand it entirely. – M.B Jul 15 '12 at 5:20
@Florian: I think that M.B. does intend that $\sigma$ and $\tau$ should range over all skew-symmetric $n \times n$ matrices. – Geoff Robinson Jul 15 '12 at 10:36
@Florian: I am so thankful to you for your comments. Indeed I did my computations for $n\geq 3$, and I thought this should work for $n=2$, which based on your comment is wrong. I don't know how to work with Gap. Would you please tell me how I can checked it for $n=3$? By the way, as Prof. Robinson mentioned here I take all skew-symmetric matrices. – M.B Jul 15 '12 at 15:56
@M.B: You can in principle use the GAP code I just posted to check for arbitrary $n$ and $p$, but I see no reason why the matrices should generate the group for bigger choices of $n$ and $p$. – Florian Eisele Jul 15 '12 at 16:47

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.