Stable reflexive sheaves and localization (Q1684744)

The paper studies the moduli space \(\mathcal{N}(2,c_1,c_2,c_3)\) of rank 2 \(\mu\)-stable reflexive sheaves on \(\mathbb{P}^3\). The action of the torus \(T\) on \(\mathbb{P}^3\) lifts to \(\mathcal{N}(2,c_1,c_2,c_3)\). With the first and second Chern classes fixed, the authors compute the generating function \(Z_{c_1,c_2}^{refl}(q):=\sum_{c_3}e(\mathcal{N}(2,c_1,c_2,c_3))q^{c_3}\) by counting the \(T\)-fixed points in \(\mathcal{N}(2,c_1,c_2,c_3)\). Every \(T\)-equivariant rank 2 reflexive sheaf on \(\mathbb{P}^3\) can be described by its toric data \[(\mathbf{u},\mathbf{v},\mathbf{p}):=\{(u_i,v_i,p_i)\}_{i=1,2,3,4},\] where \(u_i\in\mathbb{Z},v_i\in\mathbb{Z}_{\geq 0}\) and \(p_i\in Gr(1,2)\cong\mathbb{P}^2\). Every \(T\)-fixed sheaf has a unique \(T\)-equivariant structure with \(u_1=u_2=u_3=0\) in the toric data. Therefore the number of \(T\)-fixed points in \(\mathcal{N}(2,c_1,c_2,c_3)\) can be obtained by counting the toric data with extra condition \(u_1=u_2=u_3=0\) which turns out to be a combinatorial problem. The authors classify \(T\)-equivariant rank 2 reflexive sheaves into three types by their toric data. The final formula of \(Z_{c_1,c_2}^{refl}(q)\) is given by counting sheaves of each type and adding the numbers together. Another result of the paper is that the third Chern class \(c_3\) of any rank 2 reflexive sheaf must be non-negative and must have an upper bound given by \(c_1\) and \(c_2\). Moreover, the authors reprove the Hartshorne inequality for \(T\)-equivariant cases, which gives explicit upper bounds for \(c_3\). Those results imply that \(Z_{c_1,c_2}^{refl}(q)\) is a polynomial. Let \(\mathcal{M}(2,c_1,c_2,c_3)\) be the moduli space of rank 2 \(\mu\)-stable sheaves. If one wants to deduce the generating function \(Z_{c_1}(p,q):=\sum_{c_2,c_3}e(\mathcal{M}(2,c_1,c_2,c_3))p^{c_2}q^{c_3}\) from \(Z_{c_1}^{refl}(p,q):=\sum_{c_2}Z_{c_1,c_2}^{refl}(q)p^{c_2}\), one need study the Euler numbers of some Quot-schemes. The paper deals with the case \(c_1=-1,c_2=1\) and gives a closed formula. For a general case, it could be very complicated. Finally, the authors discuss briefly the generalization of the technics of this paper to any smooth projective toric 3-fold. Also, they look at the wall-crossing phenomena. But this is a short chapter and no concrete result is provided.