An arXiv preprint presents a general proof of the Gross–Joyce–Tanaka wall-crossing formula for equivariant enumerative invariants in Calabi–Yau four categories. The theorem is stated for any two stability conditions in the specified space, although it remains conditional on the paper’s stability, finiteness and properness assumptions.
That question is about how the formal invariants compare when the stability condition changes. In the same broad framework, the authors also treat stable pairs with fixed determinant, giving canonical classes and an analogous wall-crossing formula in the relevant Lie algebra under modified assumptions.
A construction built around virtual classes
The proof’s main technical method is a construction of well-behaved Calabi–Yau four pullback virtual classes using Jouanolou devices. These are the classes carried into the paper’s localization and invariant arguments.
Under the stated smoothness and resolution-property conditions, the resulting virtual classes are independent of which Jouanolou device is chosen and satisfy equivariant localization. The localization statement is part of the same conditional construction, rather than an empirical test of the formula.
The setting is a theoretical Calabi–Yau four abelian category, with a possible torus action, together with its moduli constructions. “Equivariant” here means that this action remains part of the virtual-class and localization framework; the paper is not analyzing an empirical population.
The invariants no longer depend on one setup choice
One central theorem constructs canonically defined semistable classes in the relevant Lie algebra. For a general reader, these are formal mathematical outputs of the moduli construction; the Lie-algebra language supplies the bracket structure used by the wall-crossing identities.
A separate result shows that the semistable classes are independent of the chosen framing functor. Changing that functor within the paper’s admissible setup therefore does not alter the classes delivered by the theorem.
The general wall-crossing statement then relates the invariants for any two stability conditions in W. Its scope is not tied to one particular pair of conditions, but it is still bounded by the assumptions that make the relevant moduli problem finite and proper.
The fixed-determinant extension follows the same formal pattern for stable pairs. Under its modified assumptions, it supplies canonical classes and an analogous wall-crossing theorem in the relevant Lie algebra, rather than a result for an experimental or observational sample.
Supporting identities sharpen the picture
The preprint also proves a flag-pushforward relation, a formula for carrying a virtual class from a flag bundle to the space beneath it. The relation applies when the flag bundle has constant Euler characteristic and includes the equivariant Euler class of its relative tangent bundle.
At a simple wall, the corresponding flag invariant on one side is related to the invariant on the other side by a sum of Lie-bracket terms. The terms are indexed by the class splittings associated with that wall, so the local identity records how the pieces combine algebraically.
The flag relations are supporting pieces of the formal framework: they organize virtual classes and class splittings within the wall-crossing argument. Their use remains subject to the hypotheses attached to the relevant propositions.
A theorem with explicit boundaries
The conclusions are conditional throughout. Depending on the statement, the required hypotheses include orientations and stability conditions, as well as smoothness, equivariance, finiteness, properness and resolution properties; the fixed-determinant pair result comes with its own modified assumptions.
That matters because the analyzed object is a formal category and its moduli constructions, not an empirical data set. The paper therefore establishes the stated identities only within the Calabi–Yau four settings covered by its assumptions; it does not automatically extend them to categories or moduli problems outside that scope.
The analogous Quot-scheme formula is not completed here. It is described as a future application that will require the relevant assumptions to be addressed.
Paper data and sources
Original title: Wall-crossing for equivariant DT4 invariants
Authors: Arkadij Bojko, Nikolas Kuhn, Henry Liu, Felix Thimm
Journal/Repository: arXiv
Status: Preprint, not yet peer-reviewed
First online: 2026-08-24
DOI: Not available
Original paper · Full text