A virtual structure for symplectic Higgs bundles

Basic Information

• Paper ID: 2510.24531
• Title: A virtual structure for symplectic Higgs bundles
• Author: Simon Schirren
• Classification: math.AG (Algebraic Geometry)
• Publication Date: November 10, 2025 (arXiv v2)
• Paper Link: https://arxiv.org/abs/2510.24531

Abstract

This paper proposes a virtual structure for the moduli space of symplectic Higgs pairs $(E,\phi)$ on a projective surface $S$. The key innovation is imposing a minimality assumption on the Chern character $\text{ch}(E)$, which forces all $E$ to be locally free. This is potentially significant for defining virtual counts and $Sp(r)$-Vafa-Witten invariants.

Research Background and Motivation

Core Problem

This paper aims to construct a perfect obstruction theory for Higgs bundles with gauge group $Sp(r)$ (symplectic group), thereby defining the corresponding Vafa-Witten invariants.

Problem Significance

1. Physical Motivation: Vafa and Witten, in studying supersymmetric Yang-Mills equations, discussed the Euler characteristic of instanton moduli spaces on real 4-dimensional manifolds. The mathematical characterization of these solutions is closely related to the moduli space of Higgs sheaves on complex surfaces.
2. Existing Progress: Tanaka and Thomas have established enumerative geometry theory for Higgs sheaf moduli spaces with gauge groups $U(r)$ and $SU(r)$, and defined the corresponding Vafa-Witten invariants.
3. S-Duality: From the perspective of S-duality, one needs to define similar invariants for other gauge groups, particularly the orthogonal and symplectic groups $O(r)$ and $Sp(r)$.

Limitations of Existing Methods

1. Gauge Group Restrictions: Existing theory focuses primarily on $U(r)$ and $SU(r)$ cases, lacking systematic treatment of other classical Lie groups.
2. Technical Obstacles: To view bundles with gauge group $O(r)$ or $Sp(r)$ as fixed points of the moduli space, one must handle the duality operation $E \mapsto E^*$, which requires sheaves to be reflexive (equivalent to locally free on smooth surfaces).
3. Stability Issues: One must ensure that the duality operation preserves stability, which does not always hold for arbitrary Chern characters.

Research Motivation

By introducing an involution $\iota: (E,\phi) \mapsto (E^*,-\phi^*)$ on the $U(r)$-Higgs bundle moduli space $N$, one can identify $Sp(r)$ and $O(r)$-Higgs bundles as fixed points, thereby utilizing equivariant localization techniques to construct virtual structures.

Core Contributions

1. Construction of Involution Action: Defines a well-defined involution $\iota: (E,\phi) \mapsto (E^*,-\phi^*)$ on the Higgs bundle moduli space $N$, and proves that under appropriate stability assumptions, this involution preserves stability.
2. Characterization of Fixed Point Loci: Proves that the fixed point locus $N^\iota$ of the involution decomposes into two branches $N_{O(r)}$ and $N_{Sp(r)}$, corresponding to orthogonal and symplectic Higgs bundles respectively.
3. Lifting of Spectral Theory Involution: Lifts the involution action to the spectral sheaf $E_\phi$ (defined on the Calabi-Yau threefold $X = \text{Tot}(K_S)$), providing an explicit formula: $E_{-\phi^*} = \text{Ext}^1(\sigma^*E_\phi, \pi^*K_S^{-1})$ where $\sigma$ is the $(-1)$-map on the fibers.
4. Equivariant Obstruction Theory: Constructs an $\iota$-equivariant Atiyah class, proving that the involution action on the virtual tangent bundle is compatible with the involution action on the moduli space tangent bundle via the Atiyah class.
5. Perfect Obstruction Theory: Through equivariant localization, establishes a perfect obstruction theory on $N_{Sp(r)}$, endowing it with a virtual fundamental class $[N_{Sp(r)}]^{\text{vir}}$ (Proposition 9.3.1).
6. Minimality Assumption: Proves that under the assumption that the Chern character $\text{ch}(E) = (r, 0, c_2)$ has minimal degree, all Higgs pairs $(E,\phi)$ in the moduli space are locally free (Claim 2.7.1).

Detailed Methodology

Task Definition

Construct a virtual structure for the moduli space $N_{Sp(r)}$ of symplectic Higgs bundles $(E,\omega,\phi)$ of rank $r$ on a projective surface $S$, where:

• $E$ is a rank $r$ vector bundle
• $\omega: E \xrightarrow{\sim} E^*$ is a skew-symmetric non-degenerate pairing
• $\phi \in \Gamma(\text{sp}(E) \otimes K_S)$ is a $K_S$-twisted symmetric endomorphism

Overall Strategy

Step One: Establishing the Involution Action

1. Involution on Individual Sheaves (Section 3.1): Define $(E,\phi) \mapsto (E^*,-\phi^*)$ and prove that this map preserves $\mu$-stability (Claim 3.1.2).
2. Involution on Universal Families (Sections 3.2-3.3): For the universal family $(E,\Phi)$ on $S \times N$, the functor $\iota: (E,\Phi) \mapsto (E^*,-\Phi^*)$ preserves flatness, inducing a classifying map $\iota: N \to N$ satisfying $\iota^2 = \text{id}$.

Step Two: Characterizing Fixed Points

A fixed point $(E,\phi) \in N^\iota$ corresponds to an isomorphism $f: E \xrightarrow{\sim} E^*$ making the following diagram commute:

E ----f----> E^*
|            |
φ|            |-φ^*
↓            ↓
E⊗K_S --f⊗1-> E^*⊗K_S

Decomposing $f$ into symmetric part $q$ and skew-symmetric part $\omega$:

• If $q \neq 0$, then $(E,q,\phi)$ is an orthogonal Higgs bundle ($\phi \in \Gamma(\text{so}(E) \otimes K_S)$)
• If $q = 0$, then $(E,\omega,\phi)$ is a symplectic Higgs bundle ($\phi \in \Gamma(\text{sp}(E) \otimes K_S)$)

This gives the decomposition $N^\iota = N_{O(r)} \sqcup N_{Sp(r)}$ (Proposition 4.3.1).

Step Three: Involution on Spectral Sheaves

The key technical innovation is decomposing the involution as $\iota = D \circ \sigma$:

1. Dualization $D: (E,\phi) \mapsto (E^*,\phi^*)$
   • For spectral sheaves: $E_{\phi^*}^D := \text{Ext}^1(E_\phi, \pi^*K_S^{-1})$ (Definition 5.1.1)
   • Using the resolution (2.2.1): $0 \to \pi_*E \otimes K_S^{-1} \xrightarrow{\pi_*\phi - \tau \cdot \text{id}} \pi_*E \to E_\phi \to 0$ we obtain $E_{\phi^*}^D = \text{coker}(\pi_*\phi^* - \tau \cdot \text{id}) \otimes \pi^*K_S^{-1}$
2. $(-1)$-Map on Fibers $\sigma: (E,\phi) \mapsto (E,-\phi)$
   • For spectral sheaves: $\sigma^*E_\phi = E_{-\phi}$ (Lemma 5.2.2)
   • $\sigma: X \to X$ is given by $(s,t) \mapsto (s,-t)$

Combining these yields: $E_{-\phi^*} = \text{Ext}^1(\sigma^*E_\phi, \pi^*K_S^{-1})$

Step Four: Virtual Differential

Constructs the lifting of the involution to the deformation-obstruction complex:

1. Lifting of Dualization (Section 7.1):
   • The virtual tangent bundle is represented as $R\mathcal{H}om_{pX}(E,E) \cong E^{\vee,\bullet} \otimes^L E^\bullet$
   • The dual complex $E^{D,\bullet} = E^{\bullet,\vee} \otimes \pi^*K_S^{-1}$
   • The dual action is implemented by swapping factors: $E^{\vee,\bullet} \otimes^L E^\bullet \to E^\bullet \otimes^L E^{\vee,\bullet}, \quad a \otimes b \mapsto -b \otimes a$
2. Lifting of $\sigma$ (Section 7.2):
   • Uses the isomorphism $E \cong \sigma^*E$ (Definition 6.2.1)
   • Induces $\sigma^*: R\mathcal{H}om_{pX}(E,E)[1] \to R\mathcal{H}om_{pX}(\sigma^*E,\sigma^*E)[1]$
3. Equivariance (Section 8): Proves that the Atiyah class $\text{At}_{E,N}: T_N \to R\mathcal{H}om_{pX}(E,E)[1]$ is $\iota$-equivariant, i.e., there exists a commutative diagram (8.1.2) such that the virtual differential is compatible with the differential on the moduli space via the Atiyah class.

Step Five: Equivariant Localization

1. Equivariant Embedding (Section 9.1): Embed $N$ into a smooth ambient space $A$ in a manner compatible with $\iota$.
2. Equivariant Presentation (Section 9.2): The $U(r)$-perfect obstruction theory $V^\bullet = [V^{-1} \to V^0] \xrightarrow{\psi} L_N$ can be chosen to be an $\iota$-equivariant two-term complex.
3. Restriction to Fixed Points (Section 9.3): Restrict $V^\bullet$ to $N_{Sp(r)} \subset N^\iota$, decomposing as $V^\bullet|_{N_{Sp(r)}} = V^{\bullet,\iota} \oplus V^{\bullet,\text{mov}}$ where $V^{\bullet,\iota}$ is the invariant part. The composition $V^{\bullet,\iota} \xrightarrow{\psi^\iota} L_N|_{N_{Sp(r)}}^\iota \to L_{N_{Sp(r)}}$ defines a perfect obstruction theory on $N_{Sp(r)}$ (Proposition 9.3.1).

Technical Innovations

1. Characterization of Involution on Spectral Correspondence: Provides for the first time an explicit formula for the duality operation of Higgs bundles at the level of spectral sheaves, translating geometric operations (dualization, sign changes) into functorial operations on spectral sheaves.
2. Decomposition Technique for Involution: Decomposes $\iota = D \circ \sigma$ into two controllable operations, handling dualization and sign changes separately, thereby simplifying technical difficulties.
3. Equivariance at the Virtual Level: Not only establishes the involution action at the moduli space level, but also lifts it to the virtual tangent bundle and proves compatibility with the Atiyah class—this is crucial for applying equivariant localization.
4. Use of Minimality Assumption: By requiring $c_2(E)$ to have minimal degree, cleverly avoids technical difficulties in handling non-locally-free sheaves, making the duality operation $E \mapsto E^*$ globally well-defined.

Experimental Setup

This is a pure mathematics theory paper with no numerical experiments. The theoretical framework is established through rigorous mathematical proofs.

Theoretical Verification Methods

1. Stability Verification: Verifies that the involution preserves stability through comparison of Hilbert polynomials (Lemma 2.6.5) and slope calculations (Claim 3.1.2).
2. Equivariance Verification: Verifies the equivariance of the Atiyah class through commutativity of diagrams (7.2.2) and (8.1.2).
3. Local Freeness Verification: Proves that the minimality assumption implies local freeness through comparison of Chern characters (Claim 2.7.1).

Experimental Results

Main Theoretical Results

Theorem 1 (Fixed Point Decomposition, Proposition 4.3.1): Under appropriate stability assumptions, the fixed point locus of the involution $\iota$ decomposes as $N^\iota = N_{O(r)} \sqcup N_{Sp(r)}$ where:

• $N_{O(r)} = N^\iota \cap \{\tilde{q} \neq 0\}$ (orthogonal Higgs bundles)
• $N_{Sp(r)} = N^\iota \cap \{\tilde{q} = 0\}$ (symplectic Higgs bundles)

Theorem 2 (Spectral Correspondence, Proposition 6.3.2): The spectral sheaf of $(E^*,-\phi^*)$ is given by: $E_{-\phi^*} = \text{Ext}^1(\sigma^*E_\phi, \pi^*K_S^{-1})$

Theorem 3 (Equivariant Atiyah Class, Proposition 8.1.7): The Atiyah class $\text{At}_{E,N}: T_N \to R\mathcal{H}om_{pX}(E,E)[1]$ is $\iota$-equivariant.

Theorem 4 (Main Result, Proposition 9.3.1): The composition map $V^{\bullet,\iota} \xrightarrow{\psi^\iota} L_N|_{N_{Sp(r)}}^\iota \to L_{N_{Sp(r)}}$ defines a perfect obstruction theory on $N_{Sp(r)}$, endowing it with a virtual fundamental class $[N_{Sp(r)}]^{\text{vir}}$.

Key Lemmas

1. Lemma 2.5.2: On a smooth surface, $E$ is reflexive if and only if it is locally free.
2. Lemma 5.1.2: If $E$ is locally free, then $R\mathcal{H}om(E_\phi, \pi^*K_S^{-1})[1] \cong E_\phi^D$
3. Lemma 8.1.4: The liftings of dualization and $\sigma$ commute: $D_* \circ \sigma_* = \sigma_* \circ D_*$

Theoretical Findings

1. Symmetry between Orthogonal and Symplectic: Orthogonal Higgs bundles correspond to $\phi \in \Gamma(\text{so}(E) \otimes K_S)$ (skew-symmetric), while symplectic Higgs bundles correspond to $\phi \in \Gamma(\text{sp}(E) \otimes K_S)$ (symmetric)—this is opposite to the symmetry of the pairing (Corollaries 10.3.4 and 10.3.6).
2. Determinant Constraint: The pairing $E \xrightarrow{\sim} E^*$ implies that $\det(E)$ is 2-torsion (Section 4.3), so one must restrict to the branch where $\det(E) \cong \mathcal{O}_S$.
3. Special Case: $Sp(2) \cong SL(2)$, so $N_{Sp(2)}$ already has a virtual structure given by Tanaka-Thomas theory (Section 1.3).

Related Work

Vafa-Witten Invariant Theory

1. VW Vafa-Witten (1994): Original physics work discussing solutions to supersymmetric Yang-Mills equations and S-duality.
2. TT1,TT2 Tanaka-Thomas (2017,2020):
   • Establish mathematical theory for gauge groups $U(r)$ and $SU(r)$
   • Define Vafa-Witten invariants as $\text{VW}^{N_{SU(r)}} = \int_{[N_{SU(r)}^{C^\times}]^{\text{vir}}} \frac{1}{e(\nu^{\text{vir}})} \in \mathbb{Q}$
   • Use $C^\times$-equivariant localization techniques
3. GK,GKL Göttsche-Kool and collaborators (2020,2024): Compute $SU(r)$-Vafa-Witten invariants for specific surfaces, establishing connections with modular forms.
4. O Oberdieck (2024): Curve counting on Enriques surfaces and the Klemm-Mariño formula.

Orthosymplectic Geometry

1. GS Gómez-Sols (2005): General theory of principal bundle moduli spaces.
2. OT Oh-Thomas (2023): Define SO(r)-orientation (Definition 2.4), counting sheaves on Calabi-Yau fourfolds.
3. B Bu (2025): Recent work on orthosymplectic Donaldson-Thomas theory using different methods (involving derived moduli and different stability conditions).

Technical Tools

1. GP Graber-Pandharipande (1999): Equivariant localization of virtual classes; this paper generalizes from $C^\times$ to $\mathbb{Z}/2\mathbb{Z}$.
2. BF Behrend-Fantechi (1997): Intrinsic normal cone approach to perfect obstruction theory.
3. HT Huybrechts-Thomas (2010): Deformation-obstruction theory via Atiyah and Kodaira-Spencer classes.
4. R Ricolfi (2020): Equivariant Atiyah class theory, providing technical foundation for Sections 7-8.

Positioning of This Work

This is the first systematic work establishing a virtual structure for $Sp(r)$-Higgs bundles, filling the gap in Vafa-Witten theory for orthosymplectic gauge groups. Compared to Bu's work, this paper adopts the perspective of involution fixed points, more directly leveraging existing $U(r)$ theory.

Conclusions and Discussion

Main Conclusions

1. Existence of Virtual Structure: Under the minimality assumption $\text{ch}(E) = (r,0,c_2)$, the symplectic Higgs bundle moduli space $N_{Sp(r)}$ carries a perfect obstruction theory and virtual fundamental class.
2. Effectiveness of Involution Method: By identifying $Sp(r)$ and $O(r)$-bundles as involution fixed points of the $U(r)$-moduli space, one can utilize equivariant localization techniques to inherit the virtual structure.
3. Compatibility of Spectral Theory: The involution action has natural characterizations at both the Higgs pair and spectral sheaf levels, compatible via the spectral correspondence.

Limitations

1. Minimality Assumption: Requires $c_2(E)$ to have minimal degree to ensure local freeness, excluding general torsion-free sheaves. The author acknowledges: "We are interested in generalizing to arbitrary $\text{ch}(E)$ including torsion-free sheaves" (Section 1.3).
2. Stability Issues for Orthogonal Group: The paper mentions "there are some stability issues for $O(r)$" (Section 1.3), thus focuses primarily on the $Sp(r)$ case.
3. Unproven Symmetry: While constructing the virtual structure, the paper has not proven that the virtual structure on $N_{Sp(r)}$ is symmetric (i.e., virtual dimension is zero), which is necessary for defining Vafa-Witten invariants (outlook in Section 1.3).
4. Global Existence of Universal Family: Strictly speaking, the universal family may only exist globally as a twisted universal family (Remark in Section 3.2), though the deformation-obstruction complex is always well-defined.

Future Directions

The author explicitly proposes three research directions in Section 1.3:

1. Generalization to Arbitrary Chern Characters:
   • Include torsion-free sheaves rather than only locally free sheaves
   • May require adopting Bu's approach: using derived moduli and different stability conditions
2. Proving Symmetry:
   • Prove that the virtual structure on $N_{Sp(r)}$ is symmetric
   • Make the degree of $[N_{Sp(r)}]^{\text{vir}}$ equal to zero
   • Use torus localization to define $Sp(r)$-Vafa-Witten invariants
3. Computing Specific Invariants:
   • Compute $Sp(r)$-Vafa-Witten invariants for specific surfaces (e.g., K3 surfaces, Enriques surfaces)
   • Study relationships with modular forms and string theory
   • Verify predictions from S-duality

In-Depth Evaluation

Strengths

1. Theoretical Innovation:
   • First systematic work establishing virtual structure theory for symplectic Higgs bundles
   • Novel and natural perspective using involution fixed points, fully leveraging existing $U(r)$ theory
   • Spectral sheaf level characterization of involution (Proposition 6.3.2) is a technical highlight
2. Technical Rigor:
   • Complete proof chain, from individual sheaves to universal families, from Higgs pairs to spectral sheaves, with clear logic
   • Careful verification of equivariance (Sections 7-8), particularly the compatibility of virtual differential with Atiyah class
   • Full utilization of modern algebraic geometry tools (derived categories, Atiyah classes, perfect obstruction theory)
3. Clarity of Exposition:
   • Well-structured with clear hierarchical organization from motivation to technical details
   • Appendices provide necessary background (orthosymplectic bundles, Hilbert polynomials)
   • Unified notation and clear definitions
4. Theoretical Value:
   • Lays foundation for defining $Sp(r)$-Vafa-Witten invariants
   • Advances Vafa-Witten theory toward generalization to arbitrary classical Lie groups
   • Involution method may apply to other moduli space problems

Weaknesses

1. Method Limitations:
   • Minimality assumption restricts applicability, cannot handle general torsion-free sheaves
   • Treatment of $O(r)$ is incomplete, stability issues unresolved
   • Depends on surface smoothness (Lemma 2.5.2), difficult to generalize to higher dimensions or singular settings
2. Completeness of Results:
   • Symmetry of virtual structure unproven, cannot directly define invariants
   • No concrete examples or computations; theoretical effectiveness lacks numerical verification
   • Relationship with Bu's different approach unclear (only mentioned in outlook)
3. Technical Details:
   • Proof of equivariant localization in Section 9 is relatively brief, citing author's previous work Sch
   • Global existence of universal family (twisted universal family) handled imprecisely
   • Derived geometric treatment of determinant map (Footnote 2) only referenced, not expanded
4. Literature Review:
   • Insufficient discussion of Bu's recent work (2025); comparison of advantages/disadvantages of two methods missing
   • Weak connection with physics literature; specific predictions from S-duality not detailed

Impact

1. Contribution to the Field:
   • Importance: Fills the gap in Vafa-Witten theory for orthosymplectic gauge groups
   • Methodology: Involution fixed points + equivariant localization method is universal, potentially inspiring research on other moduli spaces
   • Follow-up Research: Paves the way for computing $Sp(r)$-Vafa-Witten invariants
2. Practical Value:
   • Pure theory work with no immediate applications
   • Long-term potential for connection with physics through string theory dualities
   • Provides new tools for enumerative theory in algebraic geometry
3. Reproducibility:
   • Theoretical proofs are verifiable with clear logic
   • Depends on deep background knowledge (derived categories, perfect obstruction theory, spectral correspondence), requiring specialized training
   • Lacks concrete computational examples, difficult to apply directly to specific surfaces

Applicable Scenarios

1. Theoretical Research:
   • Algebraic geometers studying Higgs bundle moduli spaces
   • Enumerative geometers defining new Donaldson-Thomas type invariants
   • Representation theorists studying moduli spaces of classical Lie groups
2. Related Fields:
   • Mathematical Physics: Vafa-Witten theory, S-duality, gauge theory
   • Derived Algebraic Geometry: virtual structure theory for moduli spaces
   • Noncommutative Geometry: spectral correspondence, Fourier-Mukai transforms
3. Restrictions:
   • Requires projective surfaces (smoothness is crucial)
   • Chern character must satisfy minimality assumption
   • Primarily applicable to symplectic group $Sp(r)$; orthogonal group $O(r)$ requires further research

Overall Assessment

This is a technically solid and theoretically innovative algebraic geometry paper. The author cleverly utilizes the involution fixed point perspective, transforming the virtual structure problem for symplectic Higgs bundles into an equivariant localization problem, fully leveraging existing $U(r)$ theory. The spectral sheaf level characterization of involution is a technical highlight, and the rigorous proof of equivariance demonstrates the author's deep expertise.

The main limitations are the restriction of the minimality assumption and the unproven symmetry of the virtual structure, which means the theory has not yet fully achieved the goal of defining invariants. However, as foundational work, this paper establishes a solid foundation for $Sp(r)$-Vafa-Witten invariant theory and has significant theoretical value.

Recommendation: ★★★★☆ (Highly recommended for researchers in algebraic geometry/enumerative geometry)

Selected References

1. TT1 Tanaka-Thomas (2020): Vafa-Witten invariants for projective surfaces I, J. Algebraic Geom. 29, 603-668. (Main theoretical foundation of this paper)
2. GP Graber-Pandharipande (1999): Localization of virtual classes, Invent. Math. 135, 487-518. (Classical reference for equivariant localization)
3. BF Behrend-Fantechi (1997): The intrinsic normal cone, Invent. Math. 128, 45-88. (Perfect obstruction theory)
4. R Ricolfi (2020): The equivariant Atiyah class, arXiv:2003.05440. (Equivariant Atiyah class theory)
5. B Bu (2025): Orthosymplectic Donaldson-Thomas theory, arXiv:2503.20667. (Recent related work using different methods)