Refined enumerative invariants and mixed Welschinger invariants

This paper establishes the invariance of signed counts of real curves on toric surfaces under specific boundary conditions via a new relative refined tropical invariant, while demonstrating that this invariance fails when conjugate point pairs are placed in the interior.

The Gist

Imagine you are an architect trying to build a specific type of bridge (a mathematical "curve") across a river. You have a set of rules: the bridge must connect certain points, and it must be built in a specific style (either "complex" or "real").

In the world of Complex Geometry (building with imaginary numbers), things are very predictable. If you move your starting points around slightly, the number of ways you can build the bridge stays exactly the same. It's like a perfect, unchanging puzzle.

But in Real Geometry (building with real numbers, like the world we see), things are messy. If you move your starting points just a tiny bit, the number of valid bridges might suddenly change. One moment you have 5 bridges, the next you have 3. This makes it hard to count them reliably.

This paper, written by Shustin and Sinichkin, is about finding a way to make counting these "Real Bridges" stable again, even when things get complicated. They do this by using a clever trick called Tropical Geometry.

The Magic Map: Tropical Geometry

Think of Tropical Geometry as a "skeleton" or a "shadow" of the real bridge. Instead of dealing with smooth, curvy lines, tropical geometry turns everything into straight lines and sharp corners (like a stick figure drawing of the bridge).

The authors use this stick-figure map to count the bridges. The beauty of this map is that it's much easier to work with, and it often reveals hidden patterns that are invisible in the smooth version.

The Big Problem: "Interior" vs. "Boundary"

The authors discovered a specific rule that makes counting work perfectly:

• The Boundary Rule: If all the "conjugate pairs" (think of these as twin points that are mirror images of each other in the complex world) are placed on the edges (the boundary) of the construction site, the count is stable. You can move the points around, and the signed count (a special way of adding and subtracting bridges) never changes.
• The Interior Problem: If you try to put those twin points inside the construction site (the interior), the stability breaks. Even if you follow all the rules, the count will jump around depending on exactly where you put the points.

Analogy: Imagine you are arranging guests at a dinner party.

• Boundary: If all the "twin" guests sit at the head of the table (the boundary), the seating arrangement is stable. You can shuffle the chairs, and the total "vibe" of the party remains the same.
• Interior: If you put the twins in the middle of the room, moving a single chair might completely change the dynamic of the whole party. The count becomes unpredictable.

The "Refined" Invariant: The Universal Translator

The authors didn't just fix the counting; they invented a new "Universal Translator" (called a Refined Invariant).

Think of this translator as a dial with two settings:

1. Setting A (Complex World): When you turn the dial to one side, the translator tells you the total number of all possible bridges (complex curves), including the ones that don't exist in the real world.
2. Setting B (Real World): When you turn the dial to the other side, it filters out the imaginary ones and gives you the "signed count" of the real bridges.

The amazing thing is that this translator works for any setting in between. It connects the messy real world to the perfect complex world with a single, smooth formula.

Why This Matters

1. Stability: They proved that if you keep your "twin" points on the boundary, you can finally trust your count of real bridges, even for very complex shapes (high genus).
2. New Tools: They created a new mathematical tool (the refined invariant) that can calculate complex numbers and real numbers from the same source.
3. Knowing the Limits: They also proved that if you don't keep the points on the boundary, you can't expect a stable count. This tells mathematicians exactly where the rules break down, which is just as important as knowing where they work.

Summary in a Nutshell

The authors found a way to count "real" geometric shapes by turning them into "stick figures" (tropical curves). They discovered that if you keep your special "twin" points on the edge of the shape, the count is stable and predictable. They built a magical dial (the refined invariant) that can switch between counting complex shapes and real shapes, proving that these two worlds are deeply connected. However, they also warned that if you move those twins into the middle of the shape, the magic stops working, and the count becomes chaotic.

Technical Summary

Here is a detailed technical summary of the paper "Refined Enumerative Invariants and Mixed Welschinger Invariants" by Eugenii Shustin and Uriel Sinichkin.

1. Problem Statement and Motivation

The Core Problem:
In enumerative geometry, counting algebraic curves satisfying specific incidence conditions is a fundamental task. Over the complex numbers, these counts are deformation-invariant (Gromov–Witten invariants). Over the real numbers, however, the naive count of real curves is generally not invariant; it depends on the specific configuration of the point constraints.

The Welschinger Solution (Genus 0):
For genus 0 (rational curves), Welschinger introduced a signed count where curves are weighted by $(-1)^k$, where $k$ is the number of solitary real nodes. This signed count is deformation-invariant for real symplectic 4-manifolds and provides a lower bound for the number of real curves.

The Obstruction (Positive Genus):
A major open problem has been extending this invariance to positive genus ($g > 0$). It is known that the signed real count is generally not invariant under the variation of point constraints in positive genus, even for totally real point conditions.

The Gap:
Previous work by the authors (Shustin and Sinichkin, 2022) introduced a "refined tropical count" using a parameter $y$. However, this count had two limitations:

1. It was not invariant in certain settings.
2. The limit as $y \to -1$ (which should correspond to the real signed count) lacked a clear geometric interpretation in general settings.

Goal:
The paper aims to restore invariance for the signed count of real curves in arbitrary genus by imposing a specific geometric constraint: all conjugate pairs of point conditions must lie on the boundary divisors of the toric surface. The authors also seek to define a refined invariant that interpolates between the complex count ($y \to 1$) and the real signed count ($y \to -1$).

---

2. Methodology

The authors utilize Tropical Geometry as a combinatorial bridge between algebraic geometry over non-Archimedean fields and the tropical limit.

Key Technical Tools:

• Tropical Correspondence Theorems: They establish bijections between algebraic curves over a field of Puiseux series ($K$) and tropical curves in $\mathbb{R}^2$.
• Enhanced Tropical Curves: They work with "enhanced" tropical curves, which include limit curves (morphisms from $\mathbb{P}^1$ to toric surfaces) associated with the vertices of the tropical graph. This structure captures the necessary algebraic data to count curves.
• Real Structures and Conjugation: They define real tropical curves equipped with an involution $c$ (conjugation). The "real" algebraic curves correspond to "conjugation-invariant" enhancements of these tropical curves.
• Refined Multiplicities: They employ the Block-Göttsche $y$-refinement of tropical multiplicities. Instead of the standard integer multiplicity $\mu(V)$, they use the quantum integer $[ \mu(V) ]_y^-$.
• Relative Invariants: They introduce Relative Refined Tropical Invariants ($RR_y$) that allow for arbitrary tangency orders along the boundary divisors.

The "Essentially Peripheral" Condition:
The central methodological innovation is restricting the point conditions to be essentially peripheral. This means:

• Real points can be anywhere.
• Conjugate pairs of complex points must be located on the boundary divisors of the toric surface.
• The tropicalization of these pairs results in a single point with a doubled weight vector on the boundary.

---

3. Key Contributions and Results

A. Invariance of the Signed Real Count (Theorem 1.1)

The authors prove that if all conjugate pairs of point conditions are located on the boundary divisors of a real toric surface, the signed count of real curves of arbitrary genus is invariant under the variation of the points (provided the reduced tropicalization is in general position).

• Significance: This resolves the obstruction to invariance in positive genus for this specific class of constraints.
• Mechanism: The invariance is proven by showing that the refined tropical count is invariant, and its specialization at $y \to -1$ yields the signed real count.

B. The Relative Refined Tropical Invariant (Section 5)

The authors define a new invariant $RR_y(P, g, \langle u_i \rangle)$:
$$RR_y = \sum_{\Gamma} \prod_{V} [\mu(V)]_y^- \prod_{E} [wt(E)]_y^{-1}$$

• Invariance: This sum is invariant under the variation of point conditions in general position.
• Limits:
  • $y \to 1$: The invariant recovers the complex enumerative count of curves with prescribed tangency to the boundary divisors (Theorem 3.5, Prop 5.4).
  • $y \to -1$: The invariant recovers the signed count of real curves (Theorem 4.10, Prop 5.5).
• Extension: They extend this to allow arbitrary tangency orders along the boundary, linking the $y \to 1$ limit to complex curves with prescribed tangency.

C. Negative Result (Section 6)

The paper rigorously demonstrates that the invariance fails if the "essentially peripheral" condition is violated.

• Proposition 6.2 & 6.3: If conjugate pairs are allowed in the interior of the toric surface (even under strong genericity assumptions), the signed count is not invariant in positive genus.
• Example: For genus $g=1$ and degree $d=4$ in $\mathbb{P}^2$, the signed count can be 63 or 69 depending on the position of the conjugate pair, proving the necessity of the boundary condition for invariance.

D. Computational Framework

• The authors provide a method to compute these invariants using floor diagrams (Proposition 5.6) when all boundary points lie on a single boundary component (e.g., in the projective plane).
• They derive explicit formulas for the signed count based on the combinatorial structure of the tropical curve's "real" and "imaginary" subgraphs (Theorem 4.10).

---

4. Significance and Impact

1. Resolution of a Long-standing Obstruction: The paper provides the first proof of invariance for signed real curve counts in arbitrary genus, albeit under the specific "boundary" constraint. This clarifies the precise geometric conditions required for real enumerative invariance.
2. Unification of Complex and Real Counts: By introducing the refined invariant $RR_y$, the authors unify the complex count ($y=1$) and the real signed count ($y=-1$) into a single polynomial invariant. This suggests a deeper structural relationship between complex and real enumerative geometry mediated by tropical geometry.
3. Refined Tropical Geometry: The work advances the theory of refined tropical invariants, moving beyond genus 0 and providing a robust framework for handling tangency conditions and boundary constraints.
4. Negative Results as Guidance: By explicitly characterizing where invariance fails (interior conjugate pairs), the paper sets clear boundaries for future research, preventing futile attempts to find invariance in settings where it is mathematically impossible.

Summary Conclusion

Shustin and Sinichkin successfully define a refined enumerative invariant for real toric surfaces that is invariant under point variation in any genus, provided conjugate point pairs are restricted to the boundary. This invariant specializes to the classical complex count at $y=1$ and the Welschinger signed count at $y=-1$. The work establishes a rigorous tropical correspondence for these "mixed" Welschinger invariants and demonstrates that relaxing the boundary condition destroys invariance in positive genus.