Birational rigidity of complete intersections (Q510010)

Let \(X := X_{d_1,\ldots,d_s} \subset \mathbb{P}^N\) be a complete intersection of \(s\) hypersurfaces of degrees \(d_i\). Suppose also that \(X\) is a \(\mathbb{Q}\)-Fano of index \(1\) (i.e. \(\sum d_i = N\)). In the paper under review the author studies the superrigidity problem for \(X\). Namely, he introduces the notion of \textit{singularity of type \((\delta,\nu,k)\)} (Definition 0.1 in the text) and proves as his main result (Theorem 0.2) that if \(X\) has only singularities of type \((\delta,\nu,2s)\), with \(2\delta + \nu + 2 \leq \frac{2(\sum_{i=1}^s d_i + 1)}{\prod_{i=1}^s d_i} - 5s\) for some integers \(\delta\geq-1,\nu\geq 1\), then \(X\) is birationally superrigid, i.e. \(\text{Aut}(X) = \text{Bir}(X)\). As one of the corollaries the author proves that each of the smooth complete intersections \(X_{2,d} \subset \mathbb{P}^{d+2}\), \(X_{3,d} \subset \mathbb{P}^{d+3}\), \(X_{4,d} \subset \mathbb{P}^{d+4}\), \(X_{2,2,d} \subset \mathbb{P}^{d+4}\) is birationally superrigid for \(d \geq 55,82,111,246\), respectively (see Corollary 0.3). There is also an interesting relation between superrigidity of \(X\) with isolated singularities and its Tyurina numbers (cf. Corollary 0.5) and a superrigidity result for the hypersurface \(\widehat{X}\) dual to \(X\) (cf. Corollary 0.6). Few words about the proofs. The argument is based on the following crucial result (Proposition 2.1). Let \(\alpha\) an effective cycle on \(X\) of pure codimension \(k\) such that \(\alpha \sim m \cdot c_1(\mathcal{O}_X(1))^k \cap [X]\). Then \(e_S(\alpha) \leq m\) for every closed subvariety \(S \subset \mathbb{P}^N\) of dimension \(\geq ks\) not meeting \(\text{Sing}(X)\) (cf. Proposition 5 in [\textit{A.V. Pukhlikov}, J. Math. Sci., New York 115, No. 3, 2428--2436 (2003; Zbl 1071.14042)]. Here \(e_S(\alpha)\) is the Samuel multiplicity of \(\alpha\) along \(S\). Suppose now that \(X\) is not birationally superrigid. Then there is a movable linear subsystem \(\mathcal{L} \subset |-\mu K_X|\), some \(\mu \in \mathbb{Z}\), such that \(c := [\text{the canonical threshold of}\;\mathcal{L}] < 1/\mu\). Let \(Z := D_1 \cdot D_2\) be the intersection of two general members \(D_i \in \mathcal{L}\). It follows from [\textit{T. de Fernex}, Invent. Math. 192, No. 3, 533--566 (2013; Zbl 1279.14019)] and the preceding assertion that \[ \dim\{x \in Z\;| \;e_x(Z) \geq 1/c^2\} \leq \delta + 2s. \] The rest of the arguments follows the lines in [\textit{T. de Fernex}, Invent. Math. 203, No. 2, 675--680 (2016; Zbl 1441.14045)].