Defects in cohomological gauge theory and Donaldson-Thomas invariants (Q514225)

Motivated to generalize Donaldson-Thomas invariants, this paper aims to lay the foundations of a theory of divisor defects in higher dimensional topological gauge theories (cf. [\textit{M. Cirafici} et al., Nucl. Phys., B 853, No. 2, 508--605 (2011; Zbl 1229.81178); Ann. Henri Poincaré 14, No. 4, 1001--1041 (2013; Zbl 1272.81113)]). Divisor defects are introduced in \S4 as the higher dimensional generalization of surface defects of four dimensional manifolds, which is reviewed in \S2 together with physical meanings. Precisely, divisor defect is defined as follows: Since this paper mostly works on a Calabi-Yau three fold, \(X\) is taken to be a complex manifold with Kähler form \(J\) and trivial canonical bundle \(K_X=\mathcal{O}_X\). \(D\) a divisor of \(X\) locally defined by \(z=re^{i\theta}=0\). Let \(\mathcal{E}\) be a \(G\)-bundle over \(X\) may not be defined on \(D\), but have a connection \(A\) locally taking the form \[ A=\alpha d\theta+\cdots, \] near \(D\). Here \(\alpha\in L\), a Levi subgroup of \(G\) contains \(\mathbb{T}_G=\mathbf{t}/\Lambda\), where \(\mathbb{T}_G\) is the maximal torus of \(G\). Then \(\mathcal{E}\) is extended to a bundle over \(X\) whose restriction to \(D\) is a \(L\)-bundle. Hence \(F_A\) takes the form \[ F_A=2\pi\alpha\delta_D+\cdots, \] near \(D\). Here \(\delta_D\) is the Poincaré dual of \(D\). The author remarks this is rooted in the field theory description of line defects; for example when talking about (electric) Wilson lines is to integrate the gauge field over the defect, while (magnetic) 't Hooft line defects are defined by excising the line from the bulk space and specifying boundary conditions for the fields. The pair \((L,\alpha)\) determines a parabolic subgroup of \(G\) and a parabolic bundle is associated to a divisor defect. and we have \[ \mathcal{E}|_D=\bigoplus_{i=1}^r\mathcal{L}_i, \] where each \(\mathcal{L}_i\) is a line bundle. By using this decomposition, the moduli space \(\tilde{\mathcal{M}}^{(\alpha)}(L;X\setminus D)\) of critical points of the Donaldson-Uhlenbeck-Yau (DUY) equations in the complement of the divisor \(D\) is defined (\S4.3 (4.19)). Relation of DUY equations and Donaldson-Thomas invariants is explained in \S3. Where Donaldson-Thomas invariants are reviewed in the framework of cohomological Yang-Mills theory. Then to show correct choice of DUY equation is \[ (F_A-2\pi\alpha\delta_D)^{(0,2)}=0, \;(F_a-2\pi\alpha\delta_D)^{(1,1)}\wedge t\wedge t=0, \] where \(t=B+iJ\) is the complexified Kähler form. Here \(B\) is the background supergravity two form \(B\)-field. The moduli space taking DUY equation as above form is denoted by \(\mathcal{M}^\alpha(L;X)\) (\S4.3 (4.21)). In this case, the moduli space is denoted by \(\mathcal{M}^{(\alpha)}(L;X)\) or to stress the topological numbers of \(\mathcal{E}\), \(\mathcal{M}^{(\alpha)}_{n,\beta, u;r}\), \((n,-\beta,u)=(\mathrm{ch}_3(\mathcal{E},\mathrm{ch}_2(\mathcal{E})_1,c_1(\mathcal{E}))\). The author argues the cohomological gauge theory problem in the presence of a divisor operator reduces to the study of the intersection theory of \(\mathcal{M}^{(\alpha)}_{n,\beta,u;r)}\). On the other hand, by using the fact flat bundles with the prescribed singularity \(A=\alpha d\theta+\cdots\) are in one to one correspondence with stable parabolic bundles ([\textit{U. Bhosle}, Math. Ann. 293, No. 1, 177--192 (1992; Zbl 0782.14013)]), an alternative moduli space \(\mathcal{P}^{(\alpha)}_{\beta,n,u;r}\) of stable parabolic torsion free sheaves on \(X\) with a certain parabolic structure along \(D\) is introduced (\S4.5). The author says relation between these two moduli spaces is not known, but hopefully \(\mathcal{P}^{(\alpha)}_{n,\beta,u;r}\) can be seen as a compactification of \(\mathcal{M}^{\alpha)}_{\beta,n, u;r}\) in some sense. These general theory is adopted to the case \(X=\mathbb{C}^3\), in \S5 and \S6. Namely in \S5, the general instanton counting formalism for Donaldson-Thomas invariants on \(\mathbb{C}^3\) is briefly reviewed and the modifications which occur in the presence of a defect is discussed in \S6. It argue when \(X=\mathbb{C}^3\), the problem of Donaldson-Thomas theory in the presence of a divisor operator, can be reduced to counting \(\Gamma\)-equivariant instantons configualtions on \(\mathbb{C}^3\), where \(\Gamma\) is an appropirate orbifold action determined by the defect. The author says if the orbifold group and the toric group commute, and one can use virtual localization by considering the set of all torus fixed points of \(\mathcal{M}^{\mathrm{inst}}_{n,0;r}(\mathbb{C}^3)\) which are also invariant under the action of \(\Gamma\), the task studying the \(\Gamma\)-fixed point set of \(\mathcal{M}^{\mathrm{inst}}_{n,0;r}(\mathbb{C}^3)\) is greatly simplified by the knowledge of local model. Based on the introduction of a certain quiver quantum mechanism, a formalism to deal with these situations was developed in [\textit{M. Cirafici} et al., Nucl. Phys., B 853, No. 2, 508--605 (2011; Zbl 1229.81178)]. In \S7, an alternative development of this formalism is given, with some corrections of discussions of [loc. cit.]. It is done via counting generating function of Donaldson-Thomas invariant with a divisor defect (\S7. (7.24), (7.25)). Brief discussions on divisor defects in other higher dimensional cohomological field theory is given in \S8. It concludes to say \(\mathcal{M}^{\mathrm{inst}}_{\mathrm{ch}_i;r}(M_8)\) is poorly understand and technical difficulty is quite challenging. Still we believe that the geometrical problem associated the moduli group \(\mathcal{P}^\alpha(M_8;D_6)\) is rather interesting and deserve further study. Then this paper concluded to summarize findings and a discussion about possible generalizations.