Let $V$ be a representation of $G=GL_n$ (or more generally any reductive group $G$) over an algebraically closed field $\mathbb k$ of characteristic $p$. Let $[V/G]$ be the corresponding quotient stack. I was wondering what can be said about the cohomology $R\Gamma([V/G], \mathcal{O}_{[V/G]})$ (or in other words about the derived $G$-invariants of the ring of functions $\Gamma(V)$)? Namely I would like to know how much does it differ from the usual invariants $\Gamma(V)^G$ in terms of the representation $V$ (plus how bad $p$ is for $G$).

The problem is in positive characteristic $G$ is usually no more linearly reductive and I do not see any easy way to compute the higher cohomologies.

When $G$ is equal to $GL_n$ I would be particularly interested in the case when $V$ is a direct sum of copies of the tautological $n$-dimensional representation (and so the question is how $R\Gamma$ depends on $p$). In general any references/comments about that will be helpful.



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.