Exact values of generic subrank (Q7001173)

The subrank of a tensor, defined by \textit{V. Strassen} [J. Reine Angew. Math. 375/376, 406--443 (1987; Zbl 0621.68026)], plays a key role in algebraic complexity theory, particularly in the analysis of matrix multiplication complexity. As the dual counterpart to tensor rank, subrank is one of the fundamental tensor invariants and has been the subject of extensive research. In recent years, several other tensor invariants have been proposed and explored, e.g., slice rank, analytic rank, and geometric rank. The subrank serves as a lower bound for each of these invariants.\N\NThe authors study the generic values of the subrank, that is, the subrank of ``almost all'' tensors within a given tensor space. More precisely, they determine the subrank of a generic tensor in \(\mathbb{C}^{n\times n \times n}\) to be \(Q(n)=\left\lfloor {\sqrt{3n -2}} \right\rfloor \) by providing a lower bound to the known upper bound. More generally, the generic subrank of tensors of all orders and dimensions is determined. This resolves two open problems previously posed in [\textit{H. Derksen} et al., J. Lond. Math. Soc., II. Ser. 110, No. 2, Article ID e12963, 26 p. (2024; Zbl 1547.15020)]. Additionally, the authors compute the dimensions of the varieties consisting of tensors with subrank at least \(r\). Finally, some possible directions for further research are also included.