🔢 mathematics

Moduli spaces of curves with polynomial point counts

This paper establishes that the number of curves of genus $g$ over finite fields is a polynomial function of the field size if and only if $g \le 8$ , while also determining the threshold for marked points where this property fails and computing the thirteenth cohomology group of these moduli spaces as a key tool.

Original authors: Samir Canning, Hannah Larson, Sam Payne, Thomas Willwacher

Published 2026-04-21

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Samir Canning, Hannah Larson, Sam Payne, Thomas Willwacher

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine you are a mathematician trying to count the number of unique shapes you can build using a specific set of rules. In this paper, the "shapes" are algebraic curves (think of them as smooth, twisted loops or donuts with holes), and the "rules" involve the number of holes they have (called genus) and how many points are marked on them.

The authors are asking a very specific question: If we build these shapes over a "finite universe" (a mathematical world with a limited number of points, like a pixelated screen), does the number of unique shapes follow a simple, predictable pattern?

Specifically, they want to know if the count can be described by a polynomial.

A polynomial is like a simple recipe: $f(q) = q^2 + 3q + 1$ . If you plug in the size of your universe ( $q$ ), you get the answer. It's smooth, predictable, and easy to calculate.
Not a polynomial means the count is chaotic, jagged, or depends on hidden, complex structures that don't fit a simple formula.

Here is the breakdown of their discovery, using some creative analogies:

1. The "Sweet Spot" of Simplicity

The authors discovered a strict "cut-off" point for simplicity.

The Rule: If your curve has 8 or fewer holes (genus $g \le 8$ ), the number of shapes is always a simple polynomial. It's like baking cookies: no matter how big the oven is, the number of cookies follows a perfect formula.
The Break: If your curve has 9 or more holes ( $g \ge 9$ ), the pattern breaks. The count becomes chaotic. It's no longer a simple recipe; it's like trying to predict the weather with a simple equation—it just gets too complex.

Analogy: Imagine building towers with LEGO bricks.

Low Genus (1–8 holes): You can only build towers in very stable, predictable ways. The number of ways to build them grows in a smooth, predictable curve.
High Genus (9+ holes): Suddenly, the bricks start interacting in weird, unpredictable ways. You can build towers that wobble, twist, and collapse in ways that don't follow a simple math rule. The "chaos" of the extra holes ruins the pattern.

2. The "Marked Points" Twist

The paper also asks: What if we put "flags" (marked points) on these curves?
They found another rule for when the pattern breaks, depending on both the number of holes ( $g$ ) and the number of flags ( $n$ ).

The Formula: The pattern holds as long as $3g + 2n < 25$ .
The Trade-off: You can have a few flags on a simple curve, or a few holes on a curve with no flags. But if you combine too many holes and too many flags, the complexity explodes, and the simple counting formula stops working.

Analogy: Think of a dance floor.

If there are few dancers (low $g$ ) and few partners (low $n$ ), everyone can dance in a neat, predictable line.
But if you pack the floor with too many dancers and too many partners, the dance becomes a chaotic mosh pit. You can no longer predict exactly how many unique dance formations exist.

3. The "Ghost" in the Machine (Cohomology)

How did they prove this? They didn't just count shapes; they looked at the "soul" of the shapes using a tool called cohomology.

Think of cohomology as a fingerprint or a DNA test for these shapes.
The authors found that for simple shapes (low genus), the DNA is "pure" (called Tate type). It's like a clean, white noise signal that produces a smooth polynomial count.
For complex shapes (high genus), the DNA gets "contaminated" with a specific, weird signal (related to weight 11 and weight 13). This contamination is like a hidden ghost in the machine. It's a subtle, complex vibration that disrupts the simple counting pattern.

The Detective Work:
The authors had to hunt down these "ghosts."

They proved that for genus 9 to 70, a specific "ghost" (weight 11) appears and ruins the pattern.
They found a special exception: Genus 12. Here, the weight 11 ghost is silent! But, they discovered a different ghost (weight 13) that shows up there instead, ensuring the pattern still breaks.
They used powerful computers to simulate these ghosts and prove that they never disappear for high enough genus.

4. Why Does This Matter?

You might ask, "Who cares about counting shapes on finite fields?"

The Bridge: This connects two worlds: Geometry (the shape of things) and Number Theory (the properties of numbers).
The Insight: The fact that the count stops being a polynomial tells us that the underlying geometry has become too rich and complex to be described by simple algebra. It's a boundary where order turns into chaos.
The Surprise: The fact that the "chaos" starts exactly at genus 9 (and not 10, or 5) is a deep, mysterious feature of the universe of curves. It suggests there is a fundamental structural limit to how "simple" these shapes can be before they become "wild."

Summary

The paper is a mathematical detective story. The authors proved that:

Simple shapes (few holes) have a predictable, polynomial count.
Complex shapes (many holes) have a chaotic, non-polynomial count.
The transition happens at a precise mathematical boundary ( $g=9$ or $3g+2n=25$ ).
They solved the mystery by finding the hidden "ghosts" (cohomology groups) that appear exactly when the simplicity breaks down.

It's a beautiful example of how mathematicians use abstract tools to find the exact moment when order turns into complexity.

1. Problem Statement

The central problem addressed is determining for which genera $g$ and numbers of marked points $n$ the number of geometric isomorphism classes of smooth projective curves over a finite field $\mathbb{F}_q$ , denoted $N_{g,n}(q)$ , is a polynomial function of $q$ .

Context: For low genera ( $g \le 4$ ), it was known that $N_g(q)$ is polynomial. Recent work established this for $g=5,6$ .
The Question: Does there exist a threshold where this polynomiality breaks down? Specifically, for a fixed $g$ , what is the smallest $n$ such that $N_{g,n}(q)$ is not a polynomial?
Theoretical Link: By the Grothendieck–Lefschetz trace formula, $N_{g,n}(q)$ is polynomial if and only if the odd-weight Euler characteristics of the compactly supported cohomology of the moduli space $\mathcal{M}_{g,n}$ vanish. A stronger condition is that the cohomology is of Tate type (i.e., the weight filtration is supported only on even weights, and the graded pieces are sums of Tate twists).

2. Methodology

The authors employ a multi-faceted approach combining algebraic geometry, representation theory, and asymptotic analysis:

Trace Formula and Cohomology: They utilize the identification of point counts with the graded trace of Frobenius acting on $\ell$ -adic cohomology. They establish that non-polynomiality arises from the non-vanishing of odd-weight Euler characteristics, specifically $\chi_{11}$ and $\chi_{13}$ .
Getzler-Kapranov Graph Complexes: To compute cohomology, they use the Feynman transform of the modular cooperad of moduli spaces. This translates the problem into analyzing graph complexes where vertices are decorated by cohomology classes of lower-genus moduli spaces.
Symmetry and Stable Graphs: They analyze the action of symmetric groups on stable graphs. A key technical tool is the concept of "excess" (Definition 3.4), which measures the deviation of a cohomology class from being Tate. They prove that if the excess of all non-Tate components exceeds a certain bound ( $3g+2n$ ), the cohomology of the target space remains Tate.
Explicit Computation of $H^{13}$ : A major technical hurdle is the computation of the 13th cohomology group $H^{13}(\mathcal{M}_{g,n})$ . They construct explicit generators using Gysin pushforwards from boundary divisors involving $H^{11}(\mathcal{M}_{1,11})$ (related to the cusp form $\Delta$ of weight 12).
Asymptotic Analysis of Generating Functions: For large $g$ , they analyze the generating function for the weight 11 Euler characteristic $\chi_{11}(\mathcal{M}_g)$ . They derive precise asymptotic formulas using Bernoulli numbers, Riemann zeta functions, and complex analysis to show that $\chi_{11}$ grows super-exponentially and does not vanish for large $g$ .
Computer Verification: They use computer algebra (Mathematica) to verify non-vanishing of Euler characteristics for $g < 600$ and to compute specific values for small $g, n$ .

3. Key Contributions and Results

A. Main Theorems on Polynomiality

Theorem 1.1: The function $N_g(q)$ $N_{g} (q)$ is a polynomial in $q$ $q$ if and only if $g \le 8$ $g \leq 8$ .
- For $g \le 8$ , the cohomology is of Tate type.
- For $g \ge 9$ (except $g=12$ ), the weight 11 Euler characteristic $\chi_{11}(\mathcal{M}_g) \neq 0$ .
- For $g=12$ , $\chi_{11}(\mathcal{M}_{12}) = 0$ , but the weight 13 Euler characteristic $\chi_{13}(\mathcal{M}_{12}) \neq 0$ .
Theorem 1.4 (Marked Points): For a fixed genus $g \ge 1$ , the smallest integer $n$ such that $N_{g,n}(q)$ is not a polynomial is:
$n_{min} = \max\left( \left\lceil \frac{25 - 3g}{2} \right\rceil, 0 \right)$
This implies that $N_{g,n}$ is polynomial if and only if $3g + 2n < 25$ (or $g=0$ ).

B. Cohomological Results

Theorem 1.2 & 1.5: The rational cohomology of $\mathcal{M}_{g,n}$ is of Tate type if and only if $g=0$ or $3g + 2n < 25$ .
Theorem 1.7 (Computation of $H^{13}$ ): The authors provide a complete description of $H^{13}(\mathcal{M}_{g,n})$ $H^{13} (M_{g, n})$ for all $g, n$ $g, n$ .
- For $g \ge 2$ , $H^{13}(\mathcal{M}_{g,n})$ is spanned by Gysin pushforwards of classes from $H^{11}(\mathcal{M}_{1,11}) \otimes L$ (where $L$ is the Tate twist).
- They explicitly determine the $S_n$ -equivariant structure of this cohomology.
Corollary 1.8: The weight 13 Euler characteristics for the two exceptional cases are:
- $\chi_{13}(\mathcal{M}_{12}) = -6$
- $\chi_{13}(\mathcal{M}_{8,1}) = -2$
  These non-zero values prove that $N_{12}$ and $N_{8,1}$ are not polynomial, despite their weight 11 characteristics vanishing.

C. Asymptotics

Theorem 1.9: They derive the asymptotic behavior of $\chi_{11}(\mathcal{M}_g)$ as $g \to \infty$ :
$\chi_{11}(\mathcal{M}_g) \sim \begin{cases} C_{\text{ev}} \frac{(-1)^{g/2}(g-2)!}{(2\pi)^g} & g \text{ even} \\ C_{\text{odd}} \frac{(-1)^{(g-1)/2}(g-2)!}{(2\pi)^g} & g \text{ odd} \end{cases}$
where $C_{\text{ev}} \approx 12.87$ and $C_{\text{odd}} \approx 23.80$ . This growth is super-exponential but smaller than the total Euler characteristic.

4. Significance

Resolution of a Long-standing Question: The paper settles the question of when moduli spaces of curves have polynomial point counts, providing a sharp boundary ( $g \le 8$ ) that was previously unknown.
New Cohomological Computations: The explicit computation of $H^{13}(\mathcal{M}_{g,n})$ is a significant independent result. It reveals that the first non-Tate cohomology appears in weight 11 (related to the cusp form $\Delta$ ), but weight 13 is required to detect non-polynomiality in specific cases ( $g=12, n=1$ ) where weight 11 vanishes.
Methodological Advancement: The paper demonstrates the power of combining graph complex techniques (Getzler-Kapranov) with precise asymptotic analysis of generating functions. It bridges the gap between "stable" cohomology (tautological rings) and "unstable" cohomology (non-Tate classes).
Conjecture Verification: The results strongly support Conjecture 1.6, which posits that polynomiality, Tate type cohomology, and the condition $3g+2n < 25$ are equivalent. The authors verify this for all $g+n < 150$ .
Understanding Unstable Cohomology: The work clarifies the structure of unstable cohomology in moduli spaces, showing that while the stable part is Tate, the unstable part is governed by modular forms (specifically the cusp form of weight 12) and appears in specific weight and genus ranges.

In summary, this paper provides a complete characterization of the arithmetic complexity of moduli spaces of curves, linking the combinatorial properties of point counts to deep structural properties of their cohomology rings.