Let $n\geq 3$. Consider the group $GL_n(\mathbb{F}_2)$, and let $J_n=\pmatrix{1 & 1 & & \cr & 1 & 1 & & \cr & & 1 & 1 \cr & & & \ddots&\ddots}$ the Jordan cell of size $n$ (wrt to the eigenvalue $1$)
Since $GL_n(\mathbb{F}_2)$ is a simple group, it is known that it can be generated by two elements. If I understood correctly, if you pick two matrices "randomly", the two matrices will be generators, but I was curious to see if I could find two "not-too-complicated" generators.
Few tests with a CAS suggest that $GL_n(\mathbb{F}_2)$ is generated by $J_n$ and $J_n^t$ for all $n\geq 3, n\neq 4$.
I have no idea if this is true in general and, if so, how to prove it, but if true, it should certainly be well-known. I would appreciate if anyone could point me out a reference, or could provide me with a proof.
As a side question, if this is true, it would be nice to have a conceptual reason for the fact that it does not for $n=4$.
Finally, if there is a nice family of two generators for which it is easy/easier to prove it is indeed generating, I would happy to see it.
Ideally, the proofs should not use the fact that $GL_n(\mathbb{F}_2)$ is simple, if possible.
The answer is yes, and follows from the main result of the following paper.
More generally for $p$ prime and $n \geq 2$, the group $\operatorname{SL}(n,p)$ is generated by a unipotent Jordan block and its transpose, except when $(n,p) = (4,2)$.