VI-modules in nondescribing characteristic. I (Q2278140)

In this paper, the author initiates an investigation of the category of \(\mathrm{VI}\)-modules, the category of functors from \(\mathrm{VI}\) (the category of finite-dimensional \(\mathbb{F}_q\)-vector spaces and linear injections over the finite field \(\mathbb{F}_q\)) to \(k\)-modules, for \(k\) a commutative Noetherian ring. The main results require non-describing characteristic, i.e., that \(q\) be invertible in \(k\). The Noetherian hypothesis ensures that the category of finitely-generated \(\mathrm{VI}\)-modules is well-behaved, by [\textit{A. Putman} and \textit{S. V. Sam}, Duke Math. J. 166, No. 13, 2521--2598 (2017; Zbl 1408.18003); \textit{S. V. Sam} and \textit{A. Snowden}, J. Am. Math. Soc. 30, No. 1, 159--203 (2017; Zbl 1347.05010)]. The maximal subgroupoid of \(\mathrm{VI}\) is the category \(\mathrm{VB}\) of finite-dimensional \(\mathbb{F}_q\)-vector spaces and bijections; a \(\mathrm{VI}\)-module is \textit{induced} if it is induced from \(\mathrm{VB}\) (using the left adjoint to restriction). This leads to the fundamental notion of \textit{semi-induced} \(\mathrm{VI}\)-modules, namely those that admit a finite filtration such that each subquotient is induced. For a \(\mathrm{VI}\)-module \(M\), an element \(x \in M(V)\) is \textit{torsion} if there exists an inclusion \(i:V \hookrightarrow W\) such that \(M(i) (x)=0\); the torsion elements form a submodule \(\Gamma M\) and there is a total right derived functor \(\mathbb{R} \Gamma (M)\); \(M\) is said to be \textit{derived saturated} if \(\mathbb{R}\Gamma (M)=0\). The author gives the following striking homological characterization of semi-induced modules: in non-describing characteristic, a finitely-generated \(\mathrm{VI}\)-module is semi-induced if and only if it is derived saturated. This is one ingredient in the proof of the main result of the paper, the \textit{shift theorem}. For a finite-dimensional vector space \(X\), the shift functor \(\Sigma^X\) on \(\mathrm{VI}\)-modules is given by \(\Sigma ^X M (V) := M (V \oplus X)\). The shift theorem states that, for \(M\) a finitely-generated \(\mathrm{VI}\)-module in non-describing characteristic, there exists \(X\) such that \(\Sigma^X M\) is semi-induced. This is a powerful result that has important corollaries for the structure of finitely-generated \(\mathrm{VI}\)-modules, for example showing the finiteness of local cohomology and giving information on Castelnuovo-Mumford regularity. The proof of the shift theorem exploits a \textit{Harish-Chandra}-like quotient \(\overline{\Sigma}^X\) of the shift functor, which is of independent interest. This is defined by \(\overline{\Sigma}^X M (V):= (\Sigma^X M (V))/\mathbf{U}_X (V)\) for a certain unipotent \(\mathrm{VI}\)-group \(V \mapsto \mathbf{U}_X (V)\). In non-describing characteristic, this has excellent properties: it is exact and behaves well with respect to torsion, whilst being much better suited to inductive arguments than \(\Sigma^X\). The author shows that it captures the essential information to allow the proof of the shift theorem. For Part II, see [the author, J. Reine Angew. Math. 781, 187--205 (2021; Zbl 1482.20008)].