Asymptotically Z-stable bundles over projective surfaces

Abstract: We study the existence of asymptotically $Z$-stable (a.Z stable) bundles over polycyclic surfaces. Our choice of polynomial central charge is related to the existence of solutions of the deformed Hermitian--Yang--Mills equations, with vanishing $B$-field, in the large-volume limit. The main result is a technique to construct rank $3$, strictly a.Z-stable bundles as extensions of a line bundle by a $μ$-stable bundle of rank $2$. In particular, this leads to new examples of strictly a.Z-stable bundles over $\mathbb{P}^2$, the product $\mathbb{P}^1\times \mathbb{P}^1$, and the blow-up $\mathrm{Bl}_q\mathbb{P}^2$. We also present an analogue of the Hoppe criterion for the a.Z-stability of vector bundles of rank $2$, which may be of independent interest.

• The paper introduces asymptotic Z-stability as a novel criterion ensuring the existence of dHYM solutions on holomorphic bundles over projective surfaces.
• It provides explicit constructions of indecomposable rank 3 bundles and establishes a Hoppe-type cohomological criterion for rank 2 bundles.
• The results differentiate dHYM gauge theory from classical HYM stability, paving the way for new research in moduli spaces and geometric analysis.

Asymptotic Z-Stability and Holomorphic Bundles on Projective Surfaces

Introduction and Context

The paper "Asymptotically Z-stable bundles over projective surfaces" (2604.20264) addresses the intersection of gauge theory and algebraic geometry by studying stability conditions for holomorphic vector bundles on polarized algebraic surfaces. Specifically, it develops the theory of "asymptotic $Z$-stability" (a.Z-stability), a stability notion governing the existence of solutions to deformed Hermitian–Yang–Mills (dHYM) equations in the large-volume limit. These equations, motivated by string theory and special Lagrangian geometry, generalize the classical Hermitian–Yang–Mills (HYM) system and involve a polynomial "central charge" closely related to a Bridgeland-type stability.

The key novel contribution is an explicit construction of rank $3$ indecomposable, strictly a.Z-stable bundles (i.e., a.Z-stable but not slope-stable in the Mumford–Takemoto sense) on several classes of projective surfaces. The authors also present a Hoppe-type cohomological criterion for a.Z-stability of rank $2$ sheaves, enriching the toolkit for stability verification beyond the classical Mumford–Takemoto and Gieseker theories. The results subsume and extend previous constructions of decomposable strictly dHYM bundles in lower ranks, demonstrating fundamentally new classes of connections arising from the dHYM theory as compared with the classical gauge-theoretic paradigm.

Mathematical and Physical Foundations

The starting point is the $k$-deformed Hermitian–Yang–Mills equation governing unitary connections on a holomorphic vector bundle $E \to X$ over a compact Kähler manifold:

$$\left( e^{-i\varphi_k(E)} \left(k\omega \otimes \mathrm{id}_E - \frac{F_A}{2\pi}\right)^n \right) = 0,$$

where $\omega$ is the Kähler form, $F_A$ is the curvature, and $\varphi_k(E)$ is the phase angle of the complexified central charge $Z_k(E)$. In the geometric large-volume regime ($3$0), the existence of solutions is predicted to coincide with the asymptotic $3$1-stability of $3$2, a condition determined by comparing the $3$3-slopes $3$4 of $3$5 and its subsheaves:

$3$6

with $3$7 denoting polarization and $3$8. This generalizes both the Mumford–Takemoto slope stability ($3$9-stability) and Gieseker stability, recovering the former in suitable limits and identifying a finer stratification of the moduli functor in the analytic versus algebro-geometric correspondence.

Notably, it is established that a.Z-stability always implies $2$0-semistability but does not coincide with it—there exist strictly a.Z-stable bundles that are not $2$1-stable, which in analytic terms correspond to bundles supporting dHYM connections but not classical HYM solutions.

Construction of Strictly a.Z-stable Bundles

The main technical contribution is an explicit construction of indecomposable, strictly a.Z-stable bundles of rank $2$2 via iterated extensions. The strategy is to start with rank $2$3 Hartshorne–Serre bundles $2$4 constructed from zero-dimensional subschemes $2$5 and divisors $2$6 on a polycyclic surface $2$7, achieving $2$8-stability through interleaved cohomological vanishings. A rank $2$9 bundle $k$0 is then realized as an extension

$k$1

where $k$2 itself arises from an extension

$k$3

Through careful analysis of Chern class inequalities, cohomology vanishing, and the relative instability loci, a sufficient criterion for strict asymptotic $k$4-stability is produced in terms of explicit intersection numbers, existence of non-trivial extension classes, and control over the scheme-theoretic support $k$5. The conditions ensure that the only possible rank $k$6 a.Z-destabilizer of $k$7 is $k$8 itself, and that $k$9 is a.Z-stable but not slope-stable.

A Hoppe-type cohomological criterion for a.Z-stability for rank $E \to X$0 bundles is established, extending classical techniques of Okonek–Schneider–Spindler and explicit calculation to the new context of $E \to X$1-slope. Specifically, vanishing of $E \to X$2 for all $E \to X$3 such that $E \to X$4 implies a.Z-stability in rank $E \to X$5.

Examples on Canonical Surfaces

The authors provide concrete instantiations of the construction on:

• $E \to X$6: Using generic intersection points $E \to X$7 of curves of specified degrees and various choices for divisor $E \to X$8, strictly a.Z-stable, non-slope-stable, rank $E \to X$9 bundles are identified. The sharp bounds on the length of $$\left( e^{-i\varphi_k(E)} \left(k\omega \otimes \mathrm{id}_E - \frac{F_A}{2\pi}\right)^n \right) = 0,$$0 and intersection numbers are verified explicitly.
• $$\left( e^{-i\varphi_k(E)} \left(k\omega \otimes \mathrm{id}_E - \frac{F_A}{2\pi}\right)^n \right) = 0,$$1: The construction is specialized to products of projective lines, with bi-degree constraints ensuring both stability and the required cohomological vanishings.
• The blowup $$\left( e^{-i\varphi_k(E)} \left(k\omega \otimes \mathrm{id}_E - \frac{F_A}{2\pi}\right)^n \right) = 0,$$2: Calculation in the context of the exceptional divisor and its intersection data demonstrates that the theory is robust to mild singularities and birational transformations.

Each example is accompanied by explicit checks of the Cartier and Chow-theoretic conditions, as well as detailed calculations of cohomology and Riemann–Roch expressions that confirm the a.Z-stability.

Theoretical Implications and Further Directions

From a gauge-theoretic perspective, the existence of strictly a.Z-stable, indecomposable bundles demonstrates the essential non-equivalence of the dHYM and classical HYM moduli problems even in the large-volume analytic regime. This underlines the necessity of considering the central-charge–twisted stability conditions in enumerative, geometric, and string-theoretic problems, particularly those arising in mirror symmetry, stability conditions on derived categories, and the study of enumerative invariants.

At a practical level, the Hoppe-type criteria and explicit extension constructions significantly streamline the identification and study of new bundles supporting dHYM solutions. This opens the way for further enumerative work, moduli space calculations, and exploration of wall-crossing phenomena for dHYM equations and their relation to Gieseker and Bridgeland-type stability.

It is anticipated that the techniques and examples in this paper could inform generalizations to higher-dimensional varieties, moduli-theoretic questions of compactification, and deeper investigation into the role of non-classical stability conditions in derived category theory and gauge theory. Analytically, the explicit description of strictly a.Z-stable but not $$\left( e^{-i\varphi_k(E)} \left(k\omega \otimes \mathrm{id}_E - \frac{F_A}{2\pi}\right)^n \right) = 0,$$3-stable bundles should serve as useful test cases for studying singular limits, moduli functor separation, and the behavior of gauge-theoretic flows.

Conclusion

This work provides a comprehensive criterion and explicit constructions for strictly asymptotically $$\left( e^{-i\varphi_k(E)} \left(k\omega \otimes \mathrm{id}_E - \frac{F_A}{2\pi}\right)^n \right) = 0,$$4-stable, indecomposable holomorphic bundles over a range of projective surfaces. The findings clarify the relationship and distinction between $$\left( e^{-i\varphi_k(E)} \left(k\omega \otimes \mathrm{id}_E - \frac{F_A}{2\pi}\right)^n \right) = 0,$$5-stability, $$\left( e^{-i\varphi_k(E)} \left(k\omega \otimes \mathrm{id}_E - \frac{F_A}{2\pi}\right)^n \right) = 0,$$6-stability, and Gieseker stability, extending the reach of dHYM gauge theory beyond classical Yang–Mills–Higgs frameworks. The cohomological and geometric analyses, along with the provided examples, suggest new avenues for research in both the analytical and algebro-geometric aspects of stability and moduli theory for vector bundles, especially relevant for developments around geometric flows, derived categories, and string-theoretic applications.