Enumerative geometry of $K3$ surfaces

This paper presents a self-contained exposition of enumerative geometry results for K3 surfaces, confirming key conjectures by Yau–Zaslow, Göttsche, and Katz–Klemm–Vafa without requiring prior knowledge of Gromov–Witten theory.

The Gist

Imagine you are an architect trying to count the number of unique, intricate sculptures that can be built inside a specific, magical room. This room is a K3 surface. In the world of mathematics, a K3 surface is a special kind of 4-dimensional shape (though we visualize it as a 2D surface) that is perfectly smooth, has no holes, and follows very strict rules of symmetry.

The paper by Thomas Dedieu is a guidebook for counting curves (like loops of string or ribbons) that live inside these magical rooms. Specifically, it asks: "How many ways can we draw a curve of a certain shape (genus) and size (degree) inside this room?"

Here is the story of the paper, broken down into simple concepts and analogies.

1. The Problem: Counting the Unseeable

In normal geometry, if you want to count how many lines pass through two points, it's easy. But in the world of K3 surfaces, things get weird.

• The Vanishing Act: If you try to count these curves using standard math tools, the answer often comes out as zero. Why? Because the room is so flexible that you can stretch it until the curves disappear entirely.
• The Solution: Mathematicians invented a special "virtual" counting method called Gromov-Witten theory. Think of this as a magical camera that doesn't just take a photo of the curves, but counts the "shadow" or the "potential" of the curves, even if they are hidden or stretched out.

2. The First Big Discovery: The Yau-Zaslow Formula

The paper starts with a famous result: How many "simple loops" (rational curves) are there in a specific type of K3 room?

• The Analogy: Imagine a garden (the K3 surface) with a specific type of flower bed (a linear system). You want to count how many single-loop vines grow there.
• The Twist: Some vines are perfect circles, but others are knotted or have knots (singularities). In the past, mathematicians only counted the perfect ones.
• The Breakthrough: The formula discovered by Yau, Zaslow, and Beauville says: "Don't just count the vines; count the complexity of their knots!"
  • A simple knot counts as 1.
  • A complex knot counts as 2, 3, or more, depending on how tangled it is.
  • When you add them all up with these "weights," you get a magical number sequence (24, 324, 3200...) that follows a beautiful pattern found in nature (related to the number 24, which appears in string theory).

3. The Second Discovery: Curves of Any Shape

The paper then moves to harder questions: "What if the curve isn't a simple loop, but a pretzel with 2 holes? Or 3 holes?"

• The Challenge: Counting these complex shapes is like trying to count every possible way a piece of clay can be molded while passing through specific checkpoints.
• The Method: The authors use a technique called degeneration. Imagine taking a complex, wobbly sculpture and slowly melting it down until it breaks into simpler, easier-to-count pieces (like a tree of simple loops).
• The Result: They found a master formula (the Göttsche-Bryan-Leung formula) that predicts the number of these complex curves. It turns out these numbers are also connected to Modular Forms—mathematical objects that look like intricate, repeating patterns, similar to the tiles on a bathroom floor but in higher dimensions.

4. The "Ghost" Curves and BPS States

Here is where it gets really tricky. Sometimes, the "virtual" count includes things that aren't really there, or counts the same thing multiple times.

• The Multiple Cover Problem: Imagine a single ribbon lying on the floor. If you wrap a second ribbon around it exactly the same way, is that a new curve? In this math world, yes, but it counts as a "fraction" of a curve.
• The Fix: The paper discusses BPS states. Think of these as the "true" fundamental particles of the curve world. The mathematicians developed a way to filter out the "ghost" copies and the "fractional" counts to find the real, physical number of unique curves.
• The Surprise: They discovered that the number of curves depends only on the size of the room and the type of curve, not on the specific decorations of the room. It's like saying the number of possible snowflakes depends only on the temperature, not on which cloud they fell from.

5. Connecting to 3D Worlds (The Threefold Connection)

The paper concludes by linking these 2D surfaces to 3D shapes (Calabi-Yau threefolds), which are crucial in String Theory (the physics of the universe).

• The Analogy: Imagine the K3 surface is a single slice of bread. The 3D shape is the whole loaf.
• The Insight: The paper shows that if you know how to count the curves on the "slices" (the K3 surfaces), you can figure out the physics of the entire "loaf" (the 3D universe).
• The Tool: They use a concept called Noether-Lefschetz theory, which is like a map showing where the "special" slices of the loaf are located. By counting these special slices, they can solve the counting problem for the whole 3D object.

Summary: What is the Big Picture?

This paper is a bridge between pure math and theoretical physics.

1. It solves a counting puzzle: It gives us exact formulas to count complex shapes in a magical 4D room.
2. It reveals hidden patterns: The answers aren't random; they follow beautiful, repeating mathematical rhythms (modular forms).
3. It connects worlds: It shows that counting curves on a 2D surface is the key to understanding the geometry of 3D universes used in string theory.

In a nutshell: The author is teaching us how to count the invisible, tangled threads of the universe by using a special "virtual" lens, and discovering that the count follows a perfect, musical rhythm that connects geometry to the fundamental laws of physics.

Technical Summary

Technical Summary: Enumerative Geometry of K3 Surfaces

Author: Thomas Dedieu
Source: arXiv:2603.11136v1 (2026)

1. Problem Statement

The central problem addressed in this paper is the enumerative geometry of curves on K3 surfaces. Specifically, the author seeks to determine the number of curves of a given genus $g$ and homology class $\beta$ on a K3 surface $S$, often constrained by passing through a specific number of general points.

The paper tackles three distinct levels of difficulty regarding these counting problems:

1. Primitive Classes: Counting rational curves ($g=0$) in a primitive linear system $|L|$.
2. Arbitrary Genus in Primitive Classes: Counting curves of genus $g$ in a primitive class passing through $g$ points.
3. Non-Primitive Classes: Counting curves in classes $mL$ (where $m > 1$), which involves complex phenomena like multiple covers and reducible curves.

A fundamental obstacle is that standard Gromov–Witten (GW) invariants for K3 surfaces vanish because the moduli spaces of stable maps have a virtual dimension ($g-1$) that does not match the actual dimension of the family of curves ($g$). Furthermore, K3 surfaces deform to non-algebraic surfaces containing no curves at all, rendering standard deformation-invariant counts trivial.

2. Methodology

The paper employs a synthesis of classical algebraic geometry, moduli space theory, and advanced techniques from mathematical physics (mirror symmetry, BPS states).

A. Classical and Topological Approaches (Section 2)

• Compactified Jacobians: For rational curves, the author utilizes the universal compactified Jacobian $\bar{J}^p \mathcal{C}$ over the linear system $|L|$.
• Euler Characteristic Counting: The number of rational curves is derived by computing the topological Euler number $e(\bar{J}^p \mathcal{C})$. The author proves that this Euler number equals the sum of the Euler numbers of the compactified Jacobians of individual rational curves, weighted by their singularities.
• Multiplicity Interpretation: The multiplicity of a curve in the count is identified as the topological Euler number of its compactified Jacobian, which corresponds to the multiplicity of the origin in the equigeneric deformation space of the curve's singularities.

B. Reduced Gromov–Witten Theory (Sections 3 & 5)

To handle higher genus and non-primitive classes, the paper adopts Reduced Gromov–Witten theory (Maulik–Pandharipande).

• Obstruction Theory: Since the standard obstruction bundle $H^1(C, N_f)$ is non-trivial but contains no actual obstructions for K3 surfaces, a "reduced" obstruction theory is constructed. This yields a reduced virtual fundamental class $[M_{g,k}(S, \beta)]^{red}$ of the correct dimension ($2g$).
• Twistor Families: An alternative approach involves counting curves in the twistor family of the K3 surface (a 3-dimensional hyperkähler manifold), where the K3 surface appears as a fiber. This connects surface invariants to 3-fold invariants.

C. Degeneration and Modularity (Sections 3.3 & 6)

• Degeneration to Elliptic K3s: The proof of the Göttsche–Bryan–Leung formula involves degenerating a general K3 surface to an elliptic K3 surface with a section. The moduli space of stable maps decomposes into components involving multiple covers of the elliptic fibers and rational curves.
• Noether–Lefschetz Theory: The proof of the non-primitive case relies on families of lattice-polarized K3 surfaces. The author uses Noether–Lefschetz numbers (counting fibers with specific Picard lattice enhancements) to relate GW invariants of a 3-fold fibration to those of the K3 fibers.
• Modular Forms: The generating functions for these counts are shown to be quasi-modular forms (specifically related to Eisenstein series and the discriminant $\Delta$).

D. BPS State Counts and Mirror Symmetry (Sections 4 & 5)

• BPS Corrections: To handle non-primitive classes, the paper introduces BPS state numbers ($n_{g,m}$). These are derived from GW invariants by applying the Aspinwall–Morrison multiple cover formula, which corrects for the overcounting of multiple covers of lower-degree curves.
• Katz–Klemm–Vafa (KKV) Formula: The paper discusses the KKV formula, which relates GW invariants of K3 surfaces to those of Calabi–Yau threefolds fibered by K3s.
• Mirror Symmetry: The proof of the non-primitive Yau–Zaslow conjecture utilizes the STU model (a specific Calabi–Yau threefold) and mirror symmetry to compute invariants via hypergeometric solutions to Picard–Fuchs equations.

3. Key Contributions and Results

A. The Yau–Zaslow Formula (Primitive Rational Curves)

The paper provides a detailed proof of the Beauville–Yau–Zaslow formula (Theorem 2.1).

• Result: The number $N_p$ of rational curves in a primitive linear system $|L|$ of genus $p$ is given by the coefficient of $q^p$ in the expansion of:
  $$\sum_{p=0}^\infty N_p q^p = \prod_{n=1}^\infty \frac{1}{(1-q^n)^{24}}$$
• Significance: This confirms that the count is independent of the specific K3 surface and depends only on the genus. The multiplicity of a singular curve is rigorously defined via the Euler number of its compactified Jacobian.

B. The Göttsche–Bryan–Leung Formula (Arbitrary Genus)

The paper outlines the proof of the formula for counting genus $g$ curves in a primitive class passing through $g$ points (Theorem 3.5).

• Result: The generating series $F_g(q)$ for these counts is a quasi-modular form:
  $$F_g(q) = \left( \sum_{n=1}^\infty n \sigma_1(n) q^n \right)^g \prod_{m=1}^\infty \frac{1}{(1-q^m)^{24}}$$
• Methodology: The proof relies on degeneration to an elliptic K3 surface and combinatorial analysis of stable maps on the fibers.

C. Non-Primitive Classes and BPS Independence

The paper confirms the Yau–Zaslow conjecture for non-primitive classes (Theorem 4.14).

• Result: The BPS invariants $n_{0,m}^p$ (counting rational curves in class $mL$) are independent of the divisibility index $m$. Specifically:
  $$n_{0,m}^p = n_{0,1}^{m^2 p - m^2 + 1}$$
• Significance: This implies that the "true" number of rational curves in a non-primitive class is determined entirely by the geometry of a primitive class with a higher genus. This result was proven by Klemm, Maulik, Pandharipande, and Scheidegger using the STU model and the Harvey–Moore identity.

D. The Katz–Klemm–Vafa (KKV) Formula

The paper presents the KKV formula (Theorem 5.5), which generalizes the Yau–Zaslow formula to include Hodge integrals and higher genus.

• Result: The generating series for BPS invariants $r_{g,m}^p$ is given by a product formula involving modular forms, linking surface counts to 3-fold invariants.

4. Significance

1. Bridging Physics and Mathematics: The paper successfully translates physical concepts (BPS states, mirror symmetry, instanton numbers) into rigorous algebraic geometry proofs. It clarifies the mathematical meaning of "multiplicities" in curve counting, showing they arise from the geometry of moduli spaces (compactified Jacobians) and deformation theory.
2. Resolution of Vanishing Invariants: By introducing reduced Gromov–Witten invariants, the paper provides a robust framework for enumerative geometry on K3 surfaces, overcoming the triviality of standard GW invariants.
3. Universality of Counts: A profound result is that the enumerative invariants depend only on the topological and lattice-theoretic properties of the K3 surface (genus and divisibility), not on its specific complex structure.
4. Modularity: The discovery that these enumerative numbers are coefficients of modular forms (or quasi-modular forms) reveals a deep connection between the geometry of K3 surfaces and the theory of modular forms, allowing for the computation of infinite families of invariants from a few initial values.
5. Threefold Relations: The work establishes a precise dictionary between the enumerative geometry of K3 surfaces and Calabi–Yau threefolds, demonstrating that K3 invariants serve as the "building blocks" for 3-fold invariants via fibration structures.

In summary, Dedieu's notes provide a comprehensive roadmap from elementary topological counting to the sophisticated machinery of BPS states and mirror symmetry, unifying various conjectures (Yau–Zaslow, Göttsche, KKV) into a coherent theory of enumerative geometry for K3 surfaces.