Gromov-Witten invariants and membrane indices of fivefolds via the topological vertex

This paper conjectures and proves the existence of almost integer invariants governing the all-genus equivariant Gromov-Witten theory of Calabi-Yau fivefolds by establishing a vertex formalism that evaluates these invariants via the topological vertex for skeletal, locally anti-diagonal torus actions.

The Gist

Imagine you are trying to count the number of ways a tiny, invisible string can vibrate and wrap around a complex, multi-dimensional shape. In the world of theoretical physics and advanced mathematics, these shapes are called Calabi-Yau manifolds, and the counting process is known as Gromov-Witten theory.

Usually, mathematicians study these shapes in 3 dimensions (like a donut) or 4 dimensions. But this paper tackles a much stranger beast: a 5-dimensional shape.

Here is the breakdown of Yannik Schuler's paper, translated into everyday language with some creative analogies.

1. The Big Problem: Counting in 5D is Hard

Imagine you are trying to count how many different paths a hiker can take through a massive, 5-dimensional forest.

• The Forest: A Calabi-Yau fivefold (a complex geometric shape).
• The Hiker: A "membrane" (a 2D sheet, like a soap bubble) moving through the forest.
• The Count: The "Gromov-Witten invariant" is the number of ways this membrane can wrap around the forest.

The problem is that in 5 dimensions, the math gets incredibly messy. The numbers you get are often infinite series or complex fractions that don't seem to have a clear pattern. It's like trying to predict the weather in a storm where every raindrop has its own personality.

2. The Solution: The "Topological Vertex" Tool

The author introduces a powerful tool called the Topological Vertex.

• The Analogy: Think of the 5D forest as a giant Lego structure. Instead of trying to analyze the whole messy structure at once, you break it down into individual Lego bricks (vertices) and the connections between them (edges).
• The Trick: Usually, this "Lego tool" only works for 3D shapes. Schuler discovered that if the 5D forest has a specific, symmetrical structure (which he calls "skeletal" and "locally anti-diagonal"), you can shrink the problem down.
• The Magic: He shows that the complex 5D math at each "brick" actually behaves exactly like the simpler 3D math. It's as if he found a "dimensional reduction" cheat code. The 5D problem collapses into a 3D one, which we already know how to solve.

3. The "Membrane Index": Finding the Hidden Pattern

The paper's main goal is to prove a conjecture about these counts.

• The Conjecture: The messy, infinite series of numbers we get from counting membranes isn't random. It actually comes from a much simpler, "almost integer" source.
• The Analogy: Imagine you are listening to a chaotic orchestra. The noise seems random, but if you listen closely, you realize it's just a simple melody played by a few instruments, but with lots of echo and reverb.
• The Result: Schuler proves that if you strip away the "echo" (the complex denominators and fractions), you are left with a Membrane Index. This index is a clean, rational function (a fraction of polynomials) that governs the whole system.
• The "Denominator of Two": Interestingly, the paper finds that sometimes these clean numbers have a denominator of 2 (like 1/2 or 3/2). It's like finding that the universe's counting system occasionally needs to split a cookie in half to make the math work perfectly.

4. How They Did It: The "Capped Localization" Method

To prove this, the author used a technique called Capped Localization.

• The Analogy: Imagine you want to measure the traffic flow in a massive city. Instead of counting every car, you place "caps" on specific intersections (the vertices) and measure the flow only at those points.
• The Process:
  1. Decompose: Break the 5D shape into a graph of points and lines.
  2. Cap: Attach a "cap" to each point. This cap is a simplified version of the geometry that captures all the necessary information.
  3. Glue: Use a mathematical formula (the Topological Vertex) to glue these caps back together.
  4. Result: The messy global count is replaced by a neat sum of products of these simple caps.

5. Why This Matters

• Physics Connection: These 5D shapes are related to M-theory (a theory trying to unify all forces of nature). The "membranes" in the math correspond to M2-branes in physics.
• The "Almost Integer" Discovery: The paper suggests that the universe has a hidden layer of simplicity. Even though the raw data looks like a chaotic mess of fractions, the underlying "index" is surprisingly orderly (mostly integers, sometimes halves).
• Future Applications: This method opens the door to solving similar problems in even higher dimensions (7D, 9D, etc.) by using the same "shrink it down to 3D" trick.

Summary in a Nutshell

Yannik Schuler took a notoriously difficult problem in 5-dimensional geometry (counting how membranes wrap around shapes) and solved it by realizing that, under specific symmetrical conditions, the 5D problem is just a disguised 3D problem. By using a "Lego-brick" approach (the Topological Vertex), he proved that the chaotic numbers hiding in this 5D world actually follow a clean, predictable pattern called the Membrane Index. It's a bit like realizing that a complex symphony is just a simple tune played on a slightly different instrument.

Technical Summary

1. Problem Statement and Context

The paper addresses the computation of equivariant Gromov–Witten (GW) invariants for Calabi–Yau (CY) fivefolds equipped with a torus action. Specifically, it investigates the existence of "almost integer" invariants (membrane indices) that govern these series, analogous to the Gopakumar–Vafa integrality conjecture for Calabi–Yau threefolds.

• The Conjecture (Conjecture A): The author posits that the GW series of a CY fivefold $Z$ can be lifted to a rational function in the exponentiated torus weights ($q_i = e^{\epsilon_i}$). This rational function, denoted $\Omega_\beta$, represents the $\hat{A}$-genus of a mathematically conjectured moduli space of M2-branes on $Z$.
• The Challenge: Unlike CY threefolds, where the topological vertex formalism (Aganagic–Klemm–Mariño–Vafa, or AKMV) provides a powerful computational tool, the direct application of such methods to fivefolds is obstructed by the higher dimensionality and the complexity of Hodge integrals involved in the localization formulas.
• Specific Obstacles: The paper notes that general torus actions on fivefolds involve "quintuple Hodge integrals" and "rubber integrals" with four insertions, for which no closed formulas currently exist.

2. Methodology

The author develops a vertex formalism to evaluate GW invariants under specific geometric constraints. The methodology relies on three main pillars:

A. Geometric Assumptions

The proof is restricted to CY fivefolds $Z$ with a torus action satisfying:

1. Skeletal: The number of fixed points and one-dimensional orbits is finite.
2. Locally Anti-Diagonal: At every fixed point, the tangent space weight decomposition contains at least two opposite weights ($\epsilon$ and $-\epsilon$).
3. Curve Class Support: The curve classes $\beta$ are supported away from the "anti-diagonal strata" (edges where the weights are opposite).

B. Dimensional Reduction Trick

The core insight is a local dimensional reduction. Because the action is locally anti-diagonal, the fivefold geometry near a fixed point can be decomposed as $\mathbb{C}^3 \times \mathbb{C}^2$, where the torus acts anti-diagonally on the $\mathbb{C}^2$ factor.

• The author shows that the fivefold vertex weight reduces to the standard threefold topological vertex weight.
• The anti-diagonal weight of the $\mathbb{C}^2$ factor effectively plays the role of the genus counting variable ($g$) in the threefold theory. This reduction is derived using Mumford's relation for Chern classes of the Hodge bundle.

C. Capped Localization

The proof utilizes the capped localization technique (pioneered by Li–Liu–Liu–Zhou and Maulik–Oblomkov–Okounkov–Pandharipande).

• The GW invariants are decomposed into contributions from vertices and edges of the "T-diagram" (the one-skeleton of $Z$).
• Vertices: Represented by relative GW invariants of a partial compactification of $\mathbb{C}^5$ (specifically $U \times \mathbb{C}^2$, where $U$ is a blow-up of $\mathbb{P}^1 \times \mathbb{P}^1 \times \mathbb{P}^1$).
• Edges: Represented by relative GW invariants of a rank-4 vector bundle over $\mathbb{P}^1$.
• By evaluating these local terms using the reduction to the threefold vertex and rubber integrals, the author derives a closed combinatorial formula.

3. Key Contributions and Results

The Vertex Formalism (Theorem C / Corollary 1.8)

The paper establishes a formula for the disconnected GW invariants $GW^\bullet_\beta(Z, T)$ as a sum over tuples of partitions $\boldsymbol{\mu}$ decorating the half-edges of the T-diagram $\Gamma$:
$$GW^\bullet_\beta(Z, T) = \sum_{\boldsymbol{\mu}} \prod_{e \in E(\Gamma)} E(e, \boldsymbol{\mu}) \prod_{v \in V(\Gamma)} W(v, \boldsymbol{\mu})$$

• Edge Weights ($E$): Explicit monomials in $q_i$ involving the second Casimir invariant $\kappa(\mu)$ and framing/mixing parities.
• Vertex Weights ($W$): The standard AKMV topological vertex function $W_{\mu_1, \mu_2, \mu_3}(q)$, evaluated at the specific torus weight associated with the vertex.

Proof of Conjecture A (Theorem B / Corollary 1.11)

The author proves that for skeletal, locally anti-diagonal actions, the GW series lifts to a rational function $\Omega_\beta$ with coefficients in $\mathbb{Z}[1/2]$.

• Integrality: If the action is skeletal, the coefficients of $\Omega_\beta$ are integers.
• Denominators: The factor of $1/2$ arises from the square root of the virtual canonical bundle in K-theory. The paper demonstrates that $1/2$ is the "worst" denominator possible; higher powers of 2 do not appear.
• Plethystic Logarithm: A crucial technical step involves proving that the plethystic logarithm of the GW series preserves integrality within the localized K-theory ring (Appendix A).

Applications and Examples (Section 2)

The formalism is applied to several geometries:

1. Product Spaces ($X \times \mathbb{C}^2$): Recovers the standard topological vertex for threefolds, with the $\mathbb{C}^2$ weight acting as the genus parameter.
2. Strip Geometries: Generalizations of the resolved conifold.
3. Tot $\mathbb{P}^2(\mathcal{O}(-1)^{\oplus 3})$: A non-trivial example where the membrane indices exhibit denominators of 2 when the torus action becomes non-skeletal, confirming the sharpness of the integrality bounds.
4. Tot $\mathbb{P}^3(\mathcal{O}(-2)^{\oplus 2})$: Demonstrates the handling of "anti-diagonal edges" where the virtual class vanishes, allowing the formula to remain valid despite the curve class not being strictly supported away from all anti-diagonal strata.

4. Significance and Implications

• Bridging Dimensions: The paper successfully extends the powerful machinery of the topological vertex from dimension 3 to dimension 5, a significant step in understanding higher-dimensional enumerative geometry.
• M2-Brane Physics: It provides strong mathematical evidence for the physical conjecture that GW invariants of CY fivefolds count M2-branes. The derived "membrane indices" $\Omega_\beta$ are interpreted as the $\hat{A}$-genus of the M2-brane moduli space.
• Integrality Structure: It refines the understanding of integrality in higher-dimensional GW theory, showing that while full integrality (as in the GV conjecture for 3-folds) may be obstructed by a factor of $1/2$, the structure remains remarkably rigid.
• Limitations and Future Work: The author explicitly identifies the limitations: the method relies on the "locally anti-diagonal" condition to reduce the problem to 3D. Removing this assumption requires a deeper understanding of quintuple Hodge integrals. The paper also discusses connections to Nekrasov–Okounkov's K-theoretic Donaldson–Thomas theory and refined topological vertices, suggesting avenues for future research into non-anti-diagonal actions and M5-brane indices.

In summary, Schuler provides a rigorous combinatorial framework for computing GW invariants of a specific class of Calabi–Yau fivefolds, proving the existence of membrane indices and establishing a direct link between fivefold geometry and the well-understood topological vertex of threefolds via dimensional reduction.