Floer cohomology of platonic Lagrangians (Q2318839)

\textit{J. D. Evans} and \textit{Y. Lekili} [Sel. Math., New Ser. 21, No. 4, 1361--1404 (2015; Zbl 1403.53066)], introducing the notion of axial disc, computed the Floer cohomology of the Chiang Lagrangian [\textit{R. Chiang}, Int. Math. Res. Not. 2004, No. 45, 2437--2441 (2004; Zbl 1075.53080)] in \(\mathbb{CP}^3\) with itself. It shows that working over a field \(k\), this cohomology is non-zero if and only if the characteristic of \(k\) is 5. In \(\mathbb{CP}^3\), the Chiang Lagrangian is one of the Platonic Lagrangians; a family of Lagrangian \(\mathrm{SU}(2)\)-orbits \(L_C\) in a sequence of four Fano threefolds \(X_C\) parametrized by configurations \(C\) of points on the sphere. \(C\) is either of a triangle \(\Delta\), a tetrahedron \(T\), an octahedron \(O\) or an icosahedron \(I\). The respective threefolds are \(\mathbb{CP}^3\), the quadric, the threefold known as \(V_5\) and the Mukai-Umemura threefold \(V_{22}\). Chiang Lagrangian is \(L_\Delta\). Then the following Theorems are proved. {Theorem 1}. If \(HF^\ast(L_C,L_C;k)\) is nonzero over a filed \(k\) of characteristic \(p\), then \(p\) must be 5,2,2, or 2 for \(C\) equal to \(\Delta,T,O\) or \(I\), respectively. {Theorem 2}. Fix an orientation and a spin structure on each Lagrangian \(L_C\). Working over a field \(k\) of characteristic 5, 2, 2 and 2 in the four cases respectively, the Floer cohomology groups are given as \(\mathbb{Z}/2\) graded \(k\)-vector spaces by \begin{align*} HF^0(L_\Delta, L_\Delta;k)&\cong HF^1(L_\Delta,L_\Delta;k)\cong k \\ HF^0(L_T,L_T;k)&\cong HF^1(L_T,L_T;k)\cong k \\ HF^0(L_O,L_O;k)&\cong HF^1(L_O,L_O;k)\cong k^2 \\ HF^0(L_I,L_i;k)&\cong HF^1(L_I,L_i;k)\cong k \end{align*} Working over \(\mathbb{Z}\), the \(\mathbb{Z}/2\)-graded Floer cohomology rings are concentrated at degree 0 with \begin{align*} HF^0(L_\Delta, L_\Delta;\mathbb{Z})&\cong \mathbb{Z}/(5) \\ HF^0(L_T,L_T;\mathbb{Z})&\cong \mathbb{Z}/(4) \\ HF^0(L_O,L_O;\mathbb{Z})&\cong \mathbb{Z}[x]/(2, x^2+x+1)\cong \mathbb{F}_4 \\ HF^0(L_I,L_I;\mathbb{Z})&\cong \mathbb{Z}/(8). \end{align*} Here \(\mathbb{F}_4\) is the field with four elements. The author say that the results on \(L_\Delta\) were proved in [loc. cit.]. The others are new. The hardest part is computing \(HF^\ast(L_I,L_i;\mathbb{Z})\). Appendix B is devoted only to this calculation. Theorem 1 follows from: \begin{itemize} \item[1.] {Corollary 4.25}. (ii). If \(HF^\ast(L_c,L_c;k)\not=0\) over a field \(k\) of characteristic \(p\) then \(p\) must be 5 or 2 for \(C\) equal to \(\Delta\) or \(T\) respectively, and 2 or 19 for \(C=O\). For \(C=I\), \(p\) must be 2, 43, or 571. These are obtained via analyzing the closed open map \[ \mathcal{CO}^0: QH^\ast(X; R)\to HF^\ast(L,L;R). \] \item[2.] {Corollary 6.3}. If \(HF^\ast(L_C,L_C; k)\not=0\) over a field \(k\) of characteristic \(p\) then \(p\) must be 2. The index 4 count \(D\) for the octahedron is a power of 2. This is again derived by analyzing the closed open map. \item[3.] {Proposition. 4.32}. The only possible power of \(p\) for \(L_I\) in Corollary 4.25 (ii) is 2. This is obtained analyzing the antiholomorphic involution. \end{itemize} Theorem 2 follows from: \begin{itemize} \item[1.] Proposition 4.31. For a field of characteristic 2 we have isomorphisms of \(k\)-vector spaces \begin{align*} HF^0(L_O,L_O;k)&\cong HF^1(L_O,L_O;k)\cong k^2 \\ HF^0(L_I,L_I;k)&\cong BF^1(L_I,L_I;k)\cong k. \end{align*} This is proved from Proposition 4.29, asserting that after collapsing the grading of \(H^\ast(L,R)\) to \(\mathbb{Z}/2\), we have \(HF^\ast(L.L;R)\cong H^\ast(L,R)\), which is obtained analyzing the antiholomorphic involution. \item[2.] Proposition 6.4. We have an isomorphism of unital rings \[ HF^0(L_O,L_O;\mathbb{Z})\cong \mathbb{Z}[x]/(2,x^2+x+1). \] This is proved by using Corollary 6.3. \item[3.] Corollary 6.6. The self-dual cohomology ring of \(L_\Delta\) over \(\mathbb{Z}\) satisfies \[ HF^0(L_\Delta,L_\Delta;\mathbb{Z})\cong \mathbb{Z}/5, \quad HF^1(L_\Delta,L_\Delta;\mathbb{Z})=0.\] If \(k\) is a field of characteristic 5, then \(HF^0(L_\Delta,L_\Delta;k)\cong HF^1(L_\Delta,L_\Delta;k)\cong k\). This is a corollary of Proposition 6.5. The pair \(([x^3-y^3],[x^3+y^3])\) representing an equilateral triangle on the equator of \(PV\) which differs by a rotation through angle \(\pi\), is a regular value of the two-point index 4 evaluation map \(\mathrm{ev}_2:M_4\to L_\Delta^2\), with exactly two preimages. \end{itemize} Corresponding statements are given for the tetrahedron and the icosahedron (Corollaries 6.8 and 6.10). They follow from corresponding statements of Proposition 6.5 (Propositions 6.7 and 6.9). Following [loc. cit.], holomorphic discs in a complex manifold \(X\) whose boundaries lie on a totally real submanifold \(L\), which is homogeneous with respect to some group action are studied in \S2. Then, it is shown if a holomorphic disc \(u:(D,\partial D)\to (X,L)\) is of Maslov index 2, or 4 with a mild assumption, then \(u\) is axial, that is \(u(e^{i\theta}z)=R(\theta)u(z)\) (Lemmata 2.5, 2.6, cf. Definition 2.3). Platonic Lagrangians are explained in \S3. \S4 is the main part of this paper. First, an antiholomorphic involution \(\tau\) on a dense open orbit \(W_C\), built from exponentiating the conjugation on the Lie algebra \(\mathfrak{sl}(2,\mathbb{C})\cong \mathfrak{su}(2)\otimes\mathbb{C}\) is defined (\S4.3. cf. [\textit{K. Fukaya} et al., Geom. Topol. 21, No. 1, 1--106 (2017; Zbl 1359.53069)]). It extends across the compactification divisor when \(C\) is \(O\) or \(I\). When \(C\) is \(\Delta\) or \(T\), \(\tau\) cannot be defined but can be used reflect holomorphic discs. By gluing discs to their reflections we are able to reduce problems involving open holomorphic curves to closed curves and employ tools from algebraic geometry. The notion of a pole is also introduced (\S4.4). Then taking the theory of meromorphic functions, which derive their properties from their divisors, as guiding principle, several results on self-Floer cohomologies of Platonic Lagrangians as stated above, are derived. Section 5 returns to the Lagrangians themselves and constructs Heegaard splittings and Morse functions. This allows to calculate everything in the pearl complex (explained in \S3.6 and in Appendix A. cf. [\textit{P. Biran} and \textit{O. Cornea}, in: New perspectives and challenges in symplectic field theory. Dedicated to Yakov Eliashberg on the occasion of his 60th birthday. Providence, RI: American Mathematical Society (AMS). 1--44 (2009; Zbl 1185.53087)]) except the index-4 contributions (the index-4 case is studied in Appendix B). Combining the results of \S5 and the results on the closed-open maps (\S4.8 and \S6.1), Theorem 2 is obtained in \S6, the last section. This paper also contains three appendices; A: Transversallity for the pearl complex, B: Index 4 count for the icosahedron, C: Explicit representatives of the configurations.