Enumerative invariants in self-dual categories. I. Motivic invariants
Chenjing Bu
TL;DR
This work develops a motivic framework for counting self-dual objects in self-dual categories, extending classical GL(n) invariants to $O(n)$ and $Sp(2n)$ settings. It builds motivic Hall algebras and motivic Hall modules, together with Euler-form data and extension bundles, to define and analyze self-dual enumerative invariants, including wall-crossing formulas and no-pole theorems. The theory applies to self-dual quivers and to orthogonal/symplectic vector bundles and coherent-sheaf settings, with explicit algorithms and examples illustrating computations of invariants and their numerical (Euler-characteristic) realizations. The results establish a coherent structure for self-dual Donaldson--Thomas-type invariants and set the stage for future refinements, including Behrend-weighted DT invariants and broader geometric applications. Overall, the paper provides a comprehensive motivic toolkit for type B/C/D enumerative invariants, connecting categorical stability, filtrations, and duality to concrete computable invariants and wall-crossing behavior.
Abstract
In this series of papers, we propose a theory of enumerative invariants counting self-dual objects in self-dual categories. Ordinary enumerative invariants in abelian categories can be seen as invariants for the structure group $\mathrm{GL} (n)$, and our theory is an extension of this to structure groups $\mathrm{O} (n)$ and $\mathrm{Sp} (2n)$. Examples of our invariants include invariants counting principal orthogonal or symplectic bundles, and invariants counting self-dual quiver representations. In the present paper, we take the motivic approach, and define our invariants as elements in a ring of motives. We also extract numerical invariants by taking Euler characteristics of these elements. We prove wall-crossing formulae relating our invariants for different stability conditions. We also provide an explicit algorithm computing invariants for quiver representations, and we present some numerical results.
