Gromov--Witten theory beyond maximal contacts

This paper generalizes the local-relative correspondence beyond maximal contacts by identifying genus zero relative Gromov–Witten invariants of a smooth projective variety with a smooth nef divisor against specific orbifold invariants of multi-root stacks over a $\mathbb{P}^1$-bundle, thereby enabling the computation of relative invariants via absolute invariants of toric bundles.

The Gist

Imagine you are trying to count the number of ways a rubber band can stretch around a specific shape, like a donut or a sphere, but with a very strict rule: the rubber band must touch a specific line drawn on the shape a certain number of times. In the world of advanced mathematics (specifically Gromov-Witten theory), these "rubber bands" are mathematical curves, and the "lines" are divisors.

This paper, written by Yu Wang and Fenglong You, tackles a problem that has been a headache for mathematicians for a long time: How do we count these curves when they have to touch the line in multiple, complicated ways?

Here is the breakdown of their breakthrough, explained with everyday analogies.

1. The Problem: The "Too Many Rules" Puzzle

Imagine you are playing a game where you draw a path on a map.

• The Easy Version: You just need to touch the "River" (a divisor) once. This is easy to calculate.
• The Hard Version: You need to touch the River at three different spots, and at each spot, you have to touch it with a specific "intensity" (mathematicians call this contact order).

For decades, mathematicians could only solve the "Easy Version" or the "Hard Version" if the rules were extremely specific (called "maximal contact"). If you had to touch the river in a general, messy way (multiple contacts), the math became a tangled knot that was nearly impossible to untie.

2. The Old Solution: The "Local" Shortcut

Previously, there was a clever trick called the Local-Relative Correspondence.

• The Trick: Instead of trying to count the rubber bands touching the River on the original map, you could pretend the River didn't exist and instead count rubber bands floating in a special 3D space built above the map.
• The Catch: This trick only worked if the rubber band touched the River exactly once with maximum intensity. It was like having a key that only opened one specific door.

3. The New Discovery: The "Multi-Story Building"

Wang and You realized that this "Local" trick could be upgraded to handle any number of touches, not just one.

The Analogy: The Infinite Hotel
Imagine the original shape (the map) is a single-story house.

• The Old Trick: You could only move the problem to a "Local" version of this house, but it only worked for one guest (one touch).
• The New Trick: The authors built a Multi-Story Hotel (mathematically, a $\mathbb{P}^1$-bundle) right on top of the house.
  • This hotel has a "Ground Floor" (the original house).
  • It has a "Sky Floor" (infinity).
  • It has a "Special Floor" (a graph of a section).

They discovered that counting the rubber bands touching the River on the Ground Floor (the hard problem) is exactly the same as counting rubber bands floating inside this Multi-Story Hotel with a different set of rules.

Why is this better?

• The Hotel is Simpler: The geometry of this hotel is very structured (it's a "toric bundle," which is like a very orderly, grid-like building).
• The Rules are Easier: In the hotel, the "touching" rules transform into simple "orbifold" rules (think of them as special permissions for the rubber bands to spin or twist).
• The Result: A problem that was a tangled knot on the ground floor becomes a neat, solvable puzzle in the hotel.

4. The "Degeneration" Strategy: Breaking the Cake

How did they prove this? They used a technique called Degeneration.

The Analogy: Cutting a Cake
Imagine the Multi-Story Hotel is a giant cake. To understand how the rubber bands move through it, the authors imagine slicing the cake in half.

• Slice 1: The bottom half (the original house).
• Slice 2: The top half (the new hotel structure).
• The Cut: The place where they are glued together is the "River."

They proved that if you count the rubber bands in the whole cake, it's the same as counting how they behave in the two slices and how they match up at the cut. By carefully analyzing the "cut," they showed that the messy "multiple touches" on the ground floor perfectly match the "twisted touches" in the hotel.

5. The Grand Finale: Turning "Relative" into "Absolute"

The most powerful part of their paper is a "Domino Effect."

1. Step 1: They showed you can turn a problem with 2 touches into a problem with 1 touch in the hotel.
2. Step 2: They showed you can repeat this process. You can take the hotel, build another hotel on top of it, and turn 2 touches into 1 touch again.
3. Step 3: By repeating this $n$ times, you can turn a problem with $n+1$ touches (which was impossible to solve) into a problem with 0 touches (just floating in a high-dimensional space).

The Payoff:
Once you have reduced the problem to "0 touches," you are just counting rubber bands in a standard, well-understood space (a toric bundle). Mathematicians already have a "cheat sheet" (called the Mirror Theorem) to solve these standard problems instantly.

Summary

• The Problem: Counting curves that touch a boundary in complicated, multiple ways was too hard.
• The Solution: Build a "Multi-Story Hotel" (a specific geometric bundle) above the original space.
• The Magic: The complicated "multiple touches" on the ground floor are mathematically identical to "twisted touches" in the hotel.
• The Result: You can now take any complicated relative counting problem and convert it into a simple, standard counting problem that computers and humans can solve easily.

This paper essentially gives mathematicians a universal translator that turns a difficult, foreign language (relative invariants with many contacts) into a simple, native language (absolute invariants of toric bundles).

Technical Summary

Here is a detailed technical summary of the paper "Gromov–Witten Theory Beyond Maximal Contacts" by Yu Wang and Fenglong You.

1. Problem Statement

The paper addresses a fundamental difficulty in Relative Gromov–Witten (GW) Theory: the computation of genus zero relative invariants with multiple relative markings (i.e., curves with tangency conditions along a divisor $D$ at several distinct points).

• Context: Relative GW theory studies stable maps to a pair $(X, D)$ where $X$ is a smooth projective variety and $D$ is a smooth divisor. The "contact order" at a marking indicates the tangency degree with $D$.
• The Gap: While genus zero invariants with a single relative marking (maximal contact) are well-understood via the local-relative correspondence (relating them to absolute invariants of the total space of the line bundle $\mathcal{O}_X(-D)$), invariants with multiple relative markings are significantly harder to compute.
• Limitations of Existing Methods:
  • Degeneration: Standard degeneration formulas relate absolute invariants to relative ones but often leave one with complex relative invariants involving multiple markings.
  • Orbifold/Root Stacks: While relative invariants can be viewed as limits of orbifold invariants of root stacks, computing these for multiple markings is non-trivial and lacks a direct geometric reduction to simpler absolute theories.
  • Mirror Theorems: Existing relative mirror theorems (e.g., for one-point invariants) do not easily generalize to multi-point cases without complex auxiliary calculations (e.g., extended $I$-functions and non-trivial identities).

The authors ask: Can the relative GW theory of $(X, D)$ with $(n+1)$ relative markings be reduced to the absolute (or local) GW theory of a specific geometric target, thereby bypassing the complexity of multi-marking relative computations?

2. Methodology

The authors employ a combination of degeneration techniques, orbifold GW theory, and geometric compactification to establish a generalized correspondence.

A. Geometric Construction: The $\mathbb{P}^1$-Bundle

Instead of working directly with the line bundle $\mathcal{O}_X(-D)$ (which works only for maximal contact), the authors introduce a $\mathbb{P}^1$-bundle over $X$:
$$P := \mathbb{P}(\mathcal{O}_X(-D) \oplus \mathcal{O}_X)$$
They define two specific divisors on $P$:

1. $X_\infty$: The divisor at infinity (corresponding to the $\mathcal{O}_X$ summand).
2. $X_\sigma$: The graph of a section $\sigma$ of $\mathcal{O}_X(D)$, which intersects $X_\infty$ exactly along the original divisor $D$ (i.e., $X_\infty \cap X_\sigma = D$).

B. The Degeneration Scheme

The core of the proof involves degenerating the pair $(P, X_\infty + X_\sigma)$ to the normal cone of $D$ in $X$.

1. Degeneration: The total space $P$ is degenerated into a central fiber consisting of two components:
   • $X \times \mathbb{P}^1$ (containing $X_\infty$ and $X_\sigma$ as divisors).
   • $P_Y$ (a $\mathbb{P}^1$-bundle over the normal bundle of $D$).
2. Orbifold Degeneration Formula: The authors apply the degeneration formula for orbifold GW theory (specifically for multi-root stacks as defined in [TY23a]). This allows them to express the invariants of the degenerated space as a sum over "admissible graphs" (trees) connecting vertices on the two components.
3. Analysis of Degeneration Graphs: Through a rigorous analysis of the virtual cycles and contact orders (ages), they prove that for genus zero, the degeneration graph must be a star graph:
   • There is exactly one vertex on the $X \times \mathbb{P}^1$ side.
   • There are $n+1$ vertices on the $P_Y$ side.
   • The central vertex connects to all other vertices via single edges.
   • Crucially, they show that "mid-age" invariants (which complicate orbifold limits) do not appear in this specific setup.

C. Virtual Cycle Reduction

By analyzing the contributions of the vertices:

• Vertices on the $P_Y$ side (associated with the $n$ markings $p_1, \dots, p_n$) reduce to fiber classes of the bundle $P_Y \to D$. Their contributions cancel out with automorphism factors, effectively "removing" these markings from the relative count.
• The vertex on the $X \times \mathbb{P}^1$ side retains the original relative structure but with one fewer relative marking, effectively converting the problem into an absolute/local problem on the bundle.

3. Key Contributions and Results

A. The Main Theorem (Generalized Local-Relative Correspondence)

Theorem 1.3: The authors establish an identity between:

• LHS: Genus zero relative GW invariants of $(X, D)$ with $(n+1)$ relative markings with contact orders $\vec{k} = (k_0, \dots, k_n)$.
• RHS: Genus zero orbifold GW invariants of the pair $(P, X_\infty + X_\sigma)$ with $n$ orbifold markings (contact orders $(k_i, k_i)$) and one interior marking.

The identity is:
$$\pi_{rel,*}([M_{0, \vec{k}, n_0, \beta}(X, D)]^{vir}) = (-1)^{k_0} u_* \left( [M_{0, (k), n_0, \beta + \sum k_i f}(P, X_\infty + X_\sigma)]^{vir} \cap ev_0^*(X_\infty - \pi^* D) \right)$$
This effectively reduces the number of relative markings by 1 while increasing the target dimension by 1 (moving from $X$ to $P$).

B. Iterative Reduction to Toric Bundles

By repeating this process $n$ times, the authors derive Theorem 1.8:

• Genus zero relative GW invariants of $(X, D)$ with $(n+1)$ relative markings and ambient insertions are equal to absolute GW invariants of a toric bundle over $X$ of rank $2^{n+1}-1$(constructed via iterative$\mathbb{P}^1$-bundles).
• This reduces a difficult relative problem to a problem solvable by standard Mirror Theorems for Toric Bundles (e.g., [Bro14]).

C. Generalization to SNC Pairs

Theorem 1.7: The result is extended to Simple Normal Crossing (SNC) pairs $(X, D = \sum D_i)$. The correspondence holds for multi-root stacks, generalizing the local-orbifold correspondence beyond maximal contacts.

D. Computational Application: Landau–Ginzburg Potentials

The paper provides a streamlined computation for the proper Landau–Ginzburg potential (generating function of two-point relative invariants) for log Calabi–Yau pairs.

• Previously, computing these required complex extended relative mirror maps and auxiliary invariants ([You24b]).
• Using the new correspondence, the two-point relative invariants are reduced to one-point local invariants of a rank-2 bundle.
• Theorem 1.11: They recover the formula for the potential $W = x^{-1} \exp(g(y(q)))$ using the mirror theorem for toric bundles, offering a much simpler derivation.

4. Significance

1. Solving a Computational Bottleneck: The paper provides an effective algorithm to compute genus zero relative invariants with arbitrary numbers of relative markings, a problem previously considered intractable for general cases.
2. Unification of Theories: It bridges the gap between Relative, Local, and Orbifold GW theories. It shows that relative invariants are not just limits of orbifold invariants but can be exactly identified with absolute invariants of specific toric bundles.
3. Geometric Insight: It offers a geometric explanation for the local-relative correspondence. The "missing" divisor in the local theory is identified as the "divisor at infinity" ($X_\infty$) in the compactified bundle $P$, and the contact orders are encoded via the fiber class and the intersection with $X_\sigma$.
4. Applications to Mirror Symmetry: The results simplify the computation of mirror potentials for log Calabi–Yau pairs, confirming conjectures regarding open/closed duality and providing a robust framework for future calculations in enumerative geometry.
5. Reconstruction Theorem: Under the assumption that $H^*(X)$ is generated by $Pic(X)$, the paper proves that all genus zero primary relative invariants can be reconstructed from the one-point absolute invariants of the base $X$ (Corollary 1.10).

In summary, Wang and You have successfully generalized the local-relative correspondence, transforming the study of multi-marked relative GW invariants into a problem of absolute GW theory on toric bundles, thereby unlocking new computational pathways and deepening the structural understanding of relative enumerative geometry.