On cohomology rings of non-commutative Hilbert schemes and coha-modules (Q727693)

The Hilbert scheme of \(d\) points in an \(m\)-dimensional affine space has been a central topic for some time. The goal of this article is to study a non-commutative analogue of these Hlbert schemes by their cohomological properties. The Hilbert scheme of \(d\) points in \(\mathbb A^m\) parametrizes ideals of codimension \(d\) of the free polynomial algebra in \(m\) variables, so a direct generalization is to study the moduli space of left-ideals of codimension \(d\) in the free non-commutative algebra on \(m\) letters. \textit{M. Reineke} [Algebr. Represent. Theory 8, No. 4, 541--561 (2005; Zbl 1125.14006)] proved that the non-commutative Hilbert scheme posses a cell decomposition where the cells are parametrized by by \(m\)-ary trees with \(d\) nodes. Then the Chow group and the singular cohomology is a free group with basis given by the closures of the cells. This implies that a formula for the Poincaré polynomial and the Euler characteristic can be found. It is known that the non-commutative Hilbert scheme is a non-singular variety, and so the Chow group has a ring structure. Calculating the intersection product is a nontrivial task, and the basis of the Chow ring given by the cell decomposition is not well suited for computations. One main goal of this article is to provide another basis for the Chow ring so that the multiplicative structure can be studied by cohomology. In the commutative case, \textit{M. Lehn} and \textit{C. Sorger} [Duke Math. J. 110, No. 2, 345--357 (2001; Zbl 1093.14008)] showed that the cohomology ring of the Hilbert scheme of \(d\) points in the affine plane is isomorphic to the ring of class functions of the symmetric group \(S_d\). Their technique is to use results of \textit{I. Grojnowski} [Math. Res. Lett. 3, No. 2, 275--291 (1996; Zbl 0879.17011)] and \textit{H. Nakajima} [Ann. Math. (2) 145, No. 2, 379--388 (1997; Zbl 0915.14001)] where they prove that the direct sum over all \(d\) of all the cohomology groups has the structure of a vertex algebra which is isomorphic to the bosonic Fock space. In the non-commutative case, an analogue would be to give the cohomology of non-commutative Hilbert schemes a module structure over Kontsevich-Soibelman's Cohomological Hall algebra. The two main results in the article are the following: First, the non-commutative Hilbert scheme is a (commutative) moduli space, and so has a universal bundle. The author constructs a basis of the Chow group consisting of monomials in the Chern classes of the universal bundle. This gives a description of the Chow ring as a quotient of a polynomial ring. Secondly, the the author realizes the cohomology of non-commutative Hilbert schemes, which equals their Chow ring after extending scalars to the rationals, as a quotient of the Cohomological Hall algebra. The kernel of the quotient map is described explicitly. The first main result is a direct consequence of the fact that certain monomials in Chern classes, parametrized by \(m\)-ary trees, can be described as linear integer combinations of cell closures. The linear combinations have an upper uni-triangular base-change matrix. The theorem can be proved by expressing the filtration steps of the cell decomposition of Reineke as intersections of degeneracy loci and then proving that every irreducible component of this intersection has the necessary dimension. The second main result is based on the description of the module structure over the Cohomological Hall algebra. Here is where the definition of \textit{CoHa} comes in. It is short for \textit{Cohomological Hall algebra}. Then the author proves that the kernel of the quotient map from the CoHa to the Chow rings of non-commutative Hilbert schemes can be described using the CoHa-multiplication. Also note that the proof of this result depends on the Harder-Narashiman stratification. The author studies a commutative moduli of a non-commutative problem: For positive integers \(d,m\) and \(n\), for vector spaces \(V\) and \(W\) of dimension \(n\) and \(d\) respectively, \(\hat R\) is the vector space \(\text{Hom}(V,W)\oplus\text{End}(W)^m\) and \(G=\text{GL}(W)\) acting on \(\hat R\) by an explicit given action. For the stable points there exists a geometric \(G\)-quotient \(\pi:\hat{ R}^{\text{st}}\rightarrow H^{(m)}_{d,n}\) which is the object under study, called the \textit{non-commutative Hilbert Scheme}. The variety \(H^{(m)}_{d,n}\) is a \textit{framed quiver moduli space}, meaning that it can be stratified by its quiver representations. The author gives a thorough treatment of the theory of quivers, i.e. \textit{words and forests} which leads to the cell decomposition of the representations: \(H_{d,m}\) is covered by open subsets \(U_{S_\ast}\) with \(S_\ast\in\mathcal F_{d,n}\), the corresponding forest, each of which is isomorphic to an affine space of dimension \(N=(m-1)d^2+nd\). Now, certain closed subsets of the \(U_{S_\ast}\), explicitly given, are the cells of the cell decomposition. This gives a basis of the Chow group, and makes out the prerequisites of the computations. A connection between cell closures and Chern classes of the universal bundle is given, making it possible to choose a basis of the Chow group consisting of monomials in Chern classes. Then this basis is adjusted to make nice computations of the intersection products. The author considers the \textit{Cohomological Hall algebra} for the \(m\)-loop quiver, and define a module structure on the Chow rings of non-commutative Hilbert schemes. This is done explicitly with two following, explicit examples, explaining the structure. The article gives a very nice treatment of an algebaic geometric structure theory using representation theory.