Intersections of two Grassmannians in \(\mathbb{P}^9\) (Q2303068)

In this article, the authors investigate geometry of the intersection of two copies of \(\mathrm{Gr}(2,5)\) embedded in \(\mathbb{P}^9\). For \(i=1,2\), let \(p_i:\mathrm{Gr}(2,5)\hookrightarrow \mathbb{P}^{9}\) be a generic Plücker embedding and we denote by \(\mathrm{Gr}_i\) its image. Then the complete intersection \[ X=\mathrm{Gr}_1\cap \mathrm{Gr}_2 \subset \mathbb{P}^{9} \] is a Calabi-Yau threefold with Hodge numbers \(h^{1,1}(X)=1\) and \(h^{2,1}(X)=51\). They call such a Calabi-Yau threefold a \(GPK^3\) threefold after \textit{M. Gross} and \textit{S. Popescu} [Compos. Math. 127, No. 2, 169--228 (2001; Zbl 1063.14051)], \textit{G. Kapustka} [J. Lond. Math. Soc., II. Ser. 79, No. 1, 259--271 (2009; Zbl 1170.14025)], \textit{M. Kapustka} [J. Geom. Phys. 61, No. 8, 1309--1318 (2011; Zbl 1216.14038)] and \textit{A. Kanazawa} [Commun. Number Theory Phys. 6, No. 3, 661--696 (2012; Zbl 1274.14047)]. Let \(\mathbb{P}^\vee\) be the dual projective space of \(\mathbb{P}^9\). The projective dual \(\mathrm{Gr}_i^\vee \subset \mathbb{P}^\vee\) of \(\mathrm{Gr}_i\) is isomorphic to \(\mathrm{Gr}(2,5)\). Then the complete intersection \[ X=\mathrm{Gr}^\vee_1\cap \mathrm{Gr}^\vee_2 \subset \mathbb{P}^\vee \] is another \(GPK^3\) threefold. Thus \(X\) and \(Y\) are deformation equivalent Calabi-Yau threefolds, which they call \(GPK^3\) double mirrors. The pair \(X\) and \(Y\) attracted much attention in mirror symmetry, Hodge theory, and derived category because they are conjectured to share the same mirror [\textit{M. Miura}, Hibi toric varieties and mirror symmetry. University of Tokyo (Ph.D. thesis) (2013); \textit{T. Hayama} and \textit{A. Kanazawa}, ``Degenerating Hodge structure of one-parameter family of Calabi-Yau threefolds'', Asian J. Math. 25, No. 1, 31--42 (2021); KP. The main theorem of this article simply asserts that \(X\) and \(Y\) are not birational. The proof involves a detailed study of the moduli stack of \(GPK^3\) threefolds, providing a number of independently interesting results on the geometry of \(GPK^3\) threefolds. Recently it is shown that \(X\) and \(Y\) are derived equivalent [\textit{A. Kuznetsov} and \textit{A. Perry}, ``Categorical joins'', Preprint, \url{arXiv:1804.00144}]. Combining this with the main theorem, the pair \(X\) and \(Y\) gives the first example of deformation equivalent, derived equivalent, but non-birational Calabi-Yau threefolds. There are two main motivations for the main theorem. The first motivation is the birational Torelli problem for Calabi-Yau threefolds, which asks the following; let \(X\) and \(Y\) be smooth deformation equivalent Calabi-Yau threefolds such that there is an isomorphism of polarized Hodge structures between \(H^3(X,\mathbb{Z})\) and \(H^3(Y,\mathbb{Z})\) (modulo torsions), then are \(X\) and \(Y\) birational? Note that the usual Torelli problem (replacing birational by isomorphic in the above question) fails for Calabi-Yau threefolds [\textit{B. Szendröi}, Int. J. Math. 11, No. 3, 449--459 (2000; Zbl 1060.14523)], and hence the birational Torelli problem is a reasonable modification. The main theorem provides a counterexample to this problem. The second motivation is the problem of producing non-zero classes in the Grothendieck ring \(K_0(\mathrm{Var}/k)\) of \(k\)-varieties which are annihilated by a power of the class \(\mathbb{L} = [\mathbb{A}^1]\) of the affine line. This problem has been extensively studied since the first author proved in [\textit{L. A. Borisov}, J. Algebr. Geom. 27, No. 2, 203--209 (2018; Zbl 1415.14006)] that \(\mathbb{L}\) is a zero divisor in \(K_0(\mathrm{Var}/k)\) by using the Pfaffian-Grassmannian pair of Calabi-Yau threefolds (cf. [\textit{N. Martin}, C. R., Math., Acad. Sci. Paris 354, No. 9, 936--939 (2016; Zbl 1378.14009); \textit{A. Ito} et al., J. Algebr. Geom. 28, No. 2, 245--250 (2019; Zbl 1420.14019); Sel. Math., New Ser. 26, No. 3, Paper No. 38, 27 p. (2020; Zbl 1467.14051); \textit{A. Kuznetsov} and \textit{E. Shinder}, Sel. Math., New Ser. 24, No. 4, 3475--3500 (2018; Zbl 1450.11036)]). The usual method is to find a pair of derived equivalent Calabi-Yau manifold \(X_1\) and \(X_2\) and to show that \([X_1] - [X_2] \ne 0\) but \(([X_1] - [X_2])\mathbb{L}^e=0\) for some \(e \in \mathbb{N}\). The authors prove that \(GPK^3\) double mirrors give another such example. The article is well-written and successful in showing the beauty of geometry of \(GPK^3\) threefolds.