Torsion-Refined Gopakumar–Vafa Invariants

• The topic introduces invariants that capture genus, curve classes, and torsion charges on singular Calabi–Yau threefolds, refining classical Gopakumar–Vafa counts.
• It integrates discrete gauge symmetries and BPS state counting by using twisted derived categories and perverse-sheaf approaches over non-Kähler resolutions.
• Explicit computations, such as for the octic double solid, demonstrate modular bootstrap methods and non-commutative resolution techniques to resolve holomorphic ambiguities.

Torsion-refined Gopakumar–Vafa (GV) invariants are integer-valued enumerative invariants that simultaneously capture genus, curve class, and discrete torsion charge data for one-dimensional sheaves or BPS particles on singular Calabi–Yau threefolds. They generalize the classical GV invariants by incorporating the effects of torsion in the (co)homology or, equivalently, discrete gauge symmetries that arise in string and M-theory compactifications on such spaces. Their construction is motivated by advances in the physical understanding of five-dimensional BPS spectra with discrete charges, the geometry of non-Kähler small resolutions, and non-commutative crepant resolutions of singular Calabi–Yau varieties (Katz et al., 2022, Schimannek, 2021).

1. Formal Definition and Refined Structure

Let $X$ be a (compact) Calabi–Yau threefold with terminal nodal (conifold) singularities, admitting no projective crepant resolution. An analytic (non-Kähler) small resolution $\pi: \widehat X \to X$ then exists such that

$$H_2(\widehat X, \mathbb{Z}) \simeq H_2(X, \mathbb{Z}) \oplus \mathbb{Z}_N$$

for some $N \ge 2$, with the torsion summand $\mathbb{Z}_N$ generated by exceptional curves. The corresponding group $$H^3(\widehat X, \mathbb{Z})_{\mathrm{tors}} \simeq \mathbb{Z}_N$$ also encodes the Brauer class $\alpha$ (the fractional B-field) labeling twisted derived categories.

Every curve class in $H_2(\widehat X)$ can thus be decomposed as $(\beta, \ell)$ with

$$\beta \in H_2(X, \mathbb{Z}), \quad \ell \in \{0, 1, \dots, N-1\} \simeq \mathbb{Z}_N.$$

The torsion-refined GV invariant is then the integer $\pi: \widehat X \to X$0, attached to each triple $\pi: \widehat X \to X$1, defined via the perverse-sheaf approach but evaluated on the moduli space of 1-dimensional sheaves on all small resolutions, organized in the twisted derived category $\pi: \widehat X \to X$2 (Katz et al., 2022).

For genus-one–fibered Calabi–Yau threefolds, with $\pi: \widehat X \to X$3, the invariants can also be labeled $\pi: \widehat X \to X$4 for $\pi: \widehat X \to X$5, and refine the classical invariants $\pi: \widehat X \to X$6 (Schimannek, 2021).

2. Physical Interpretation: Discrete Gauge Symmetry and BPS Counting

Upon compactification of M-theory on $\pi: \widehat X \to X$7, a five-dimensional gauge symmetry

$\pi: \widehat X \to X$8

emerges, where the torsion factor encodes a discrete gauge symmetry. The electric charge lattice is $\pi: \widehat X \to X$9, so 5d BPS states arising from M2-branes wrapped on $$H_2(\widehat X, \mathbb{Z}) \simeq H_2(X, \mathbb{Z}) \oplus \mathbb{Z}_N$$0 acquire both continuous $$H_2(\widehat X, \mathbb{Z}) \simeq H_2(X, \mathbb{Z}) \oplus \mathbb{Z}_N$$1 charges and a discrete $$H_2(\widehat X, \mathbb{Z}) \simeq H_2(X, \mathbb{Z}) \oplus \mathbb{Z}_N$$2 charge $$H_2(\widehat X, \mathbb{Z}) \simeq H_2(X, \mathbb{Z}) \oplus \mathbb{Z}_N$$3.

These BPS multiplets transform as representations of $$H_2(\widehat X, \mathbb{Z}) \simeq H_2(X, \mathbb{Z}) \oplus \mathbb{Z}_N$$4; their refined multiplicities are denoted $$H_2(\widehat X, \mathbb{Z}) \simeq H_2(X, \mathbb{Z}) \oplus \mathbb{Z}_N$$5. The torsion-refined GV invariants are obtained by tracing over $$H_2(\widehat X, \mathbb{Z}) \simeq H_2(X, \mathbb{Z}) \oplus \mathbb{Z}_N$$6, giving

$$H_2(\widehat X, \mathbb{Z}) \simeq H_2(X, \mathbb{Z}) \oplus \mathbb{Z}_N$$7

with $$H_2(\widehat X, \mathbb{Z}) \simeq H_2(X, \mathbb{Z}) \oplus \mathbb{Z}_N$$8 encoding $$H_2(\widehat X, \mathbb{Z}) \simeq H_2(X, \mathbb{Z}) \oplus \mathbb{Z}_N$$9 representation content (Katz et al., 2022, Schimannek, 2021).

This structure directly ties the enumerative geometry to five-dimensional BPS state counting with discrete $N \ge 2$0 charge.

3. Topological String Partition Functions: Discrete Sectors and Expansion

The torsion-refined GV invariants serve as coefficients in the expansion of A-model topological string partition functions for each choice of B-field (non-commutative resolution). For each k-th sector ($N \ge 2$1),

$N \ge 2$2

with $N \ge 2$3 the Kähler parameters. The logarithm of the partition function expands as

$N \ge 2$4

with $N \ge 2$5 (Katz et al., 2022).

Introducing a bookkeeping variable $N \ge 2$6 for the torsion charge, a unified "master" partition function appears:

$N \ge 2$7

where specializing $N \ge 2$8 and $N \ge 2$9 recovers the sector-wise partition functions.

In the case of genus-one fibered Calabi–Yau threefolds, these expansions are organized using the Tate–Shafarevich group $\mathbb{Z}_N$0, where each choice of discrete B-field corresponds to an element of $\mathbb{Z}_N$1 (Schimannek, 2021).

4. Modular Structure and Connections to the Tate–Shafarevich Group

The stringy Kähler moduli of a generic genus-one curve of degree $\mathbb{Z}_N$2 is identified with the modular curve $\mathbb{Z}_N$3; the torsion classes $\mathbb{Z}_N$4 are conjecturally the Tate–Shafarevich group $\mathbb{Z}_N$5. For each element of $\mathbb{Z}_N$6, the topological string partition function displays distinctive modular properties.

The modular bootstrap in the A-model involves expressing degree-wise partition functions as Jacobi forms under $\mathbb{Z}_N$7. Notably, the Fricke involution

$\mathbb{Z}_N$8

interchanges the partition function for smooth genus-one fibrations with that for non-commutative resolutions, reflecting the equivalence of distinct large-volume phases within the same Tate–Shafarevich group element. This modular transformation also acts on the genus expansion and discrete charge assignment (Schimannek, 2021).

Cusps of the modular curve $\mathbb{Z}_N$9 correspond to large-volume limits, each associated with a distinct smooth or non-commutative Calabi–Yau geometry in the given torsion sector.

5. Explicit Computational Schemes and Example: The Octic Double Solid

Explicit computations of torsion-refined GV invariants proceed by:

• Modular bootstrap of partition functions for each base-degree, as Jacobi forms, with numerators fixed by known low-genus Gopakumar–Vafa numbers.
• Utilization of Higgs transitions in M-theory/F-theory to relate 5d BPS multiplets and specialize modular parameters for each Kähler cone phase.
• Direct calculation of mirror B-model free energies and holomorphic anomaly recursion at singular large-volume limits, including non-commutative (nc) MUM points.

A key example is the singular octic double solid $$H^3(\widehat X, \mathbb{Z})_{\mathrm{tors}} \simeq \mathbb{Z}_N$$0, a double cover of $$H^3(\widehat X, \mathbb{Z})_{\mathrm{tors}} \simeq \mathbb{Z}_N$$1 branched over a determinantal octic, with 84 conifold points and no Kähler resolution. A non-Kähler small resolution $$H^3(\widehat X, \mathbb{Z})_{\mathrm{tors}} \simeq \mathbb{Z}_N$$2 yields $$H^3(\widehat X, \mathbb{Z})_{\mathrm{tors}} \simeq \mathbb{Z}_N$$3. The two non-commutative resolutions are mirror to distinct large-volume limits of the dual intersection $$H^3(\widehat X, \mathbb{Z})_{\mathrm{tors}} \simeq \mathbb{Z}_N$$4, related via homological projective duality and hybrid GLSM phases (Katz et al., 2022).

Tables of $$H^3(\widehat X, \mathbb{Z})_{\mathrm{tors}} \simeq \mathbb{Z}_N$$5 for the octic (genus up to $$H^3(\widehat X, \mathbb{Z})_{\mathrm{tors}} \simeq \mathbb{Z}_N$$6) are obtained by fixing holomorphic ambiguities with conifold gap conditions, Castelnuovo bounds, and fractional constant-map corrections. Low-degree and low-genus torsion-refined invariants agree with direct sheaf-counts (e.g., $$H^3(\widehat X, \mathbb{Z})_{\mathrm{tors}} \simeq \mathbb{Z}_N$$7 for 4-tangent lines).

Examples for $$H^3(\widehat X, \mathbb{Z})_{\mathrm{tors}} \simeq \mathbb{Z}_N$$8 show explicit splitting of invariants by discrete charge. For $$H^3(\widehat X, \mathbb{Z})_{\mathrm{tors}} \simeq \mathbb{Z}_N$$9 and base degree $\alpha$0, $\alpha$1, $\alpha$2; for $\alpha$3, in base degree $\alpha$4, $\alpha$5, $\alpha$6, $\alpha$7 (Schimannek, 2021).

$\alpha$8	Torsion-Refined Invariants (Sample)	Comments
2	$\alpha$9: 3,0; $H_2(\widehat X)$0: -568,-512	$H_2(\widehat X)$1 sectors
4	$H_2(\widehat X)$2: 132,128,152	$H_2(\widehat X)$3 sectors
5	$H_2(\widehat X)$4: 90,100,125	$H_2(\widehat X)$5

6. Geometric, Physical, and Categorical Generalizations

Torsion-refined GV invariants unify several conceptual advances:

• Non-commutative resolutions (NCCR) are linked with elements $H_2(\widehat X)$6, with the twist provided by a fractional B-field or Clifford sheaf, and correspond to distinct points in the large-volume moduli.
• They extend beyond torus-fibered Calabi–Yau manifolds to double cover/Clifford sheaf cases and hybrid Landau–Ginzburg models.
• Fractional constant-map contributions to the free energy are computed via Szendrői's non-commutative Donaldson–Thomas partition function (Katz et al., 2022), resolving ambiguities in holomorphic anomaly recursion.
• A single framework encompasses both general-type (Jacobian) and Clifford-type (double cover) geometries.

A plausible implication is that the torsion-refined invariants give a universal indexed count of BPS particles carrying both continuous and discrete charges, with direct geometric and physical enumerative significance.

7. Relation to Previous and Parallel Developments

Initial physical definitions of GV invariants were based on BPS spectrum counting, while mathematical definitions via moduli spaces and perverse sheaves refined this to smooth Calabi–Yau threefolds. The physical refinement for torsion/discrete charges—proposed in the context of torus-fibered geometries—was formalized using the Tate–Shafarevich group to organize discrete B-field data and was made uniform for all $H_2(\widehat X)$7 in (Schimannek, 2021).

The framework in (Katz et al., 2022):

• Sharpened previous physical proposals by giving a uniform mathematical definition via twisted derived categories.
• Demonstrated the equivalence of NCCR and twisted derived categories $H_2(\widehat X)$8.
• Computed non-commutative constant-map contributions using Szendrői's methods.

These developments offer a unified and computable approach to refined enumerative geometry on singular Calabi–Yau threefolds, linking topological string theory, algebraic geometry, and modern representation theory of BPS states (Katz et al., 2022, Schimannek, 2021).