Microlocal condition for non-displaceability (Q2419129)

Let \(M\) be a symplectic manifold, \(A,B\) its compact subsets. \(A\) and \(B\) called non-displaceable if \(A\cap X(B)\not=\emptyset\), where \(X\) is any Hamiltonian symplectomorophism of \(M\) which is identity outside of a compact. Let \(\mathbb{T}^N=\{(z_0,\dots,z_N)\mid |z_0|=\cdots=|z_N|\}\) be the Clifford torus in \(\mathbb{CP}^N\), then non-dispaceability of \(\mathbb{T}^N\) from itself is already proved by using Floer's theory [\textit{C.-H. Cho}, Int. Math. Res. Not. 2004, No. 35, 1803--1843 (2004; Zbl 1079.53133); \textit{M. Entov} and \textit{L. Polterovich}, Compos. Math. 145, No. 3, 773--826 (2009; Zbl 1230.53080)]. In this paper, based on Kashiwara-Schapira's microlocal theory of sheaves on manifolds [\textit{M. Kashiwara} and \textit{P. Schapira}, Sheaves on manifolds. With a short history ``Les débuts de la théorie des faisceaux'' by Christian Houzel. Berlin etc.: Springer-Verlag (1990; Zbl 0709.18001)], a sufficient condition for non-displaceability of a pair of subsets in a cotangent bundle, independent of Floer's theory, is presented (Theorem 3.1). Then constructing Lagrangian correspondence between \(T^\ast \mathrm{SU}(N)\) and \(\mathbb{CP}^N\times(\mathbb{CP}^N)^{\mathrm{op}}\), where symplectic form of \((\mathbb{CP}^N)^{\mathrm{op}}\) is opposite that of \(\mathbb{CP}^N\), the following Theorem is derived as an application of Theorem 3.1. Theorem 4.1. \begin{itemize} \item[(1)] \(\mathbb{T}^N\) is non-displaceable from itself. \item[(2)] \(\mathbb{RP}^N\) is non-displaceable from itself. \item[(3)] \(\mathbb{T}^N\) and \(\mathbb{RP}^N\) are non-displaceable from each another. \end{itemize} To obtain sufficient condition (Theorem 3.1), first the category \(\mathcal{D}(X)=D(X\times\mathbb{R})/C_{\leq 0}(X)\), \(D(X\times \mathbb{R})\) is the unbounded derived category of \(\mathbb{K}\)-vector spaces on \(X\times \mathbb{R}\) and \(C_{\leq 0}(X)\) is the full subcategory of objects microsupport \(\Omega_{\leq 0}\); the closed subset of \(T^\ast(X\times\mathbb{R})\) such that \[ \Omega_{\leq 0}=\{\omega|(\omega,V)\leq 0\}, \quad V=\frac{d}{dt}, \] and \(\mathcal{D}_A(X)\), the full subcategory consisting of all \(F\in \mathcal{D}(X)\) such that \(SS(F)\cap \Omega_{>0}\subset \mathrm{Cone}(A)\), are introduced. (\S2). In \S2.2.2, from \(T_{c^*}:\mathcal{D}(X\times\mathbb{R})\to \mathcal{D}(X\times \mathbb{R})\), \(T_c(x,t)=(x,t+c)\), a natural transfunctor \(\tau_{c^*}:\mathrm{Id}\to T_{c^*}\) of endfunctors on \(\mathcal{D}_A(X)\) for any \(A\) is constructed. Using these, the sufficient condition is stated: Theorem 3.1. Suppose there exist objects \(F_i\in \mathcal{D}_{F_i}(X)\), \(i=1,2\), such that for all \(c>0\), the natural map \[ \tau_c:R\mathrm{hom}(F_1,F_2)\to R\mathrm{hom}(F_1, T_cF_2), \] is not zero. Then \(F_1\) and \(F_2\) are mutually non-displaceable. This is proved to use \(R\mathrm{hom}_{\mathcal{D}(x)}(F_1,F_2)=0\), if \(F_i\in\mathcal{D}_{A_i}(X)\), \(A_i\subset T^\ast X\) and \(A_1\cap A_2=\emptyset\) (Theorem 3.2), and existence of a collection of endmorphisms \(T_n:\mathcal{D}(X)\to \mathcal{D}(X)\), \(1\leq k\leq N\) for some \(N\) and a collection of functors \(t_k*T_{2k}\to T_{2k+1}\) and \(s_k:T_{2k+2}\to T_{2k+1}\) such that \(T_N=Id\), \(T_1(\mathcal{D}_L(X))\subset \mathcal{D}_{\Phi(L)}(X)\) and \(\mathrm{Cone}(t_k(F)), \mathrm{Cone}(s_k(F))\) are torsion sheaves for all \(k\) and \(F\in\mathcal{D}(X)\). Here \(\Phi:T^\ast X\to T^\ast X\) is a Hamiltonian symplectomorphism which is equal to identity outside of a compact and \(L\subset T^\ast X\) is a compact subset (Theorem 3.9). Let \(G=\mathrm{SU}(N)\), \(\mathfrak{g}\) its Lie algebra and \(\mathbb{CP}^N=\mathcal{O}\subset \mathfrak{g}^*\), and let \(I:M\to T^ast G\) be the inclusion, \(P:M\to \mathcal{O}^{\mathrm{op}}\times \mathcal{O}\) be the projection, and \(\Delta\) is the diagonal of \(\mathcal{O}^{\mathrm{op}}\times\mathcal{O}\). Then Theorem 4.1 follows from the facts that \(IP^{-1}\Delta\) and \(IP^{-1}(\mathbb{T}\times\mathbb{T})\), \(IP^{-1}\Delta\) and \(IP^{-1}(\mathbb{RP}^N\times\mathbb{RP}^N)\), and \(IP^{-1}(\mathbb{RP}^N\times\mathbb{RP}^N)\) and \(IP^{-1}(\mathbb{T}\times\mathbb{T})\) are non-displaceable each others (Theorem 4.3). Theorem 4.3 is proved from Theorem 3.1 assuming existence \(u_\mathcal{O}\in\mathcal{D}_{IP^{-1}\Delta}(G)\) which is not a torsion object, and that there exists a neighborhood \(U\) of the unit \(G\) such that for every \(g\in G, F\in D(G)\) such that \(F\) is supported on \(gU\) and \(R\Gamma(G,F)=0\), then \(F_{* G}u_\mathcal{O}\) is a torsion object (Proposition 4.4). Proof of Proposition 4.4 is the main body of this paper. In \S5, assuming unique existence a special element \(\mathfrak{S}\in D(G\times\mathfrak{h})\), \(\mathfrak{h}\) is the Cartan algebra of \(\mathfrak{g}\), \(u_\mathcal{O}\) is constructed by using convolution on \(\mathfrak{h}\); \(u_\mathcal{O}=I_0^{-1}(\mathfrak{S}\ast_\mathfrak{h}\gamma_L)\), \(\gamma_L=\mathbb{K}_{\{(A,t)|t+\langle A,L\rangle\geq 0\}}\in D(\mathfrak{h}\times\mathbb{R})\). \S6 constructs \(\mathfrak{S}\) and proves its uniqueness (Theorem 6.1, 6.7). \S7 computes an isomorphism type of \(\mathfrak{S}|_{z\times C^o_-}\), \(C^o_-\) is the interior of \(-C~+\). The author says the computation is a version of Bott's computation of \(H_\bullet (\Omega(G))\) using Morse theory. \(\mathfrak{S}\) is a strict B-sheaf. In \S8, it is shown that any strict B-sheaf can be recovered from its restriction onto \(\mathbf{Z}\times C^o_-\), \(\mathbf{Z}\) is the center of \(G\) (Theorem 8.4). This paper also includes two Appendices; \(\mathrm{SU}(N)\) and its Lie algebra: Notations and a Couple of Lemmas (\S10), and Results from [Kashiwara and Schapira, loc. cit.] on fundamental properties of microsupport (\S11). For the entire collection see [Zbl 1412.32002].