🔢 mathematics

Equidistribution of rational points and the geometric sieve for toric varieties

This paper proves the Manin–Peyre principle for the equidistribution of rational points on smooth proper split toric varieties by refining Salberger's universal torsor method to derive effective asymptotic formulas and an effective geometric sieve.

Original authors: Zhizhong Huang

Published 2026-07-20

📖 4 min read🧠 Deep dive

CC BY 4.0

Original authors: Zhizhong Huang

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 standing in a vast, invisible city built not of brick and mortar, but of pure numbers. This is the world of arithmetic geometry, a branch of mathematics where shapes (called varieties) are defined by equations, and the "people" living in them are rational points—solutions to those equations that can be written as simple fractions. For decades, mathematicians have been trying to understand how these fractional inhabitants are distributed. Do they cluster in certain neighborhoods? Do they spread out evenly across the entire city?

To make sense of this, mathematicians use a tool called a "height function." Think of height not as how tall a person is, but as a measure of how complicated a fraction is. A fraction like $1/2$ has a low height, while something like $123,456/789,012$ has a very high height. The big question is: if you look at all the rational points with a height below a certain limit (say, "all fractions smaller than a billion"), do they spread out evenly across the shape, or do they hide in specific corners? This paper tackles that question for a special, highly symmetric type of shape called a "toric variety." It's like asking if the citizens of a perfectly symmetrical, multi-dimensional crystal palace are living in a fair, balanced way throughout the whole structure.

The paper, written by Zhizhong Huang, proves that for these specific crystal palaces (smooth, proper, split toric varieties), the citizens do spread out perfectly evenly, provided you ignore a few "thin" neighborhoods where they might crowd together unnaturally. This confirms a major prediction known as the Manin–Peyre principle. But the author doesn't just stop at saying "they are evenly spread." He builds a sophisticated counting machine that tells you exactly how many people are in any given neighborhood, complete with a "margin of error" that gets smaller as the city grows larger.

Furthermore, the paper introduces a "geometric sieve." Imagine you have a giant net designed to catch rational points that accidentally land on a specific, tiny, forbidden line or dot (a subset of the shape that is too small to be a full room). The sieve proves that as you look at larger and larger heights, the number of points falling into these tiny, forbidden traps becomes negligible—almost zero. This is a powerful tool because it allows mathematicians to filter out "bad" points and focus on the "good" ones that follow the rules of the universe.

The paper also applies these findings to "fibrations," which are like layers of a cake or a stack of pancakes. If you have a shape made of many smaller shapes stacked on top of each other, the paper helps calculate how many of those layers are "everywhere locally soluble"—a fancy way of saying, "Does this layer have a solution in every possible number system (real numbers, p-adic numbers, etc.)?" The results show that if the layers are too complicated or have certain "twisted" properties, the number of fully soluble layers drops off dramatically. However, if the layers are well-behaved, the number of soluble layers grows in a predictable, beautiful pattern.

In short, this paper is a masterclass in counting. It takes a complex, high-dimensional problem about the distribution of fractions on geometric shapes and solves it with precision. It proves that these points distribute themselves according to a specific, natural measure (the Tamagawa measure), provides formulas to count them in any region with a known error margin, and demonstrates how to effectively sieve out points that fall into lower-dimensional traps. It turns a vague intuition about "even distribution" into a rigorous, calculable reality for a wide class of geometric shapes.

Technical Summary: Equidistribution of Rational Points and the Geometric Sieve for Toric Varieties

1. Problem Statement and Context

The paper addresses the Manin–Peyre principle, a conjecture concerning the asymptotic distribution of rational points of bounded height on algebraic varieties. Specifically, it investigates whether rational points on a smooth proper split toric variety $X$ over $\mathbb{Q}$ (with globally generated anticanonical line bundle) are equidistributed in the adelic space $X(\mathbb{A}_{\mathbb{Q}})$ with respect to the Tamagawa measure, provided one excludes a "thin" subset of points.

While Manin's conjecture (counting the total number of points) has been established for toric varieties via harmonic analysis (Batyrev–Tschinkel) and the universal torsor method (Salberger), the equidistribution principle (Principle 1.1) and its extension to open subvarieties (Principle 1.3, or "purity of equidistribution") have remained largely unproven for general varieties, with only rare exceptions known.

The paper aims to:

Prove the equidistribution of rational points for smooth proper split toric varieties.
Establish an effective version of this equidistribution, providing asymptotic formulas with explicit error terms dependent on the adelic neighborhood.
Prove an effective geometric sieve for these varieties, showing that rational points specializing into codimension $\ge 2$ subvarieties modulo large primes are negligible.
Apply these results to count rational points satisfying local conditions arising from fibrations and to estimate $S$ -integral points.

2. Methodology

The core methodology relies on a refinement of the universal torsor method (Salberger), combined with analytic number theory techniques and a novel "toric van der Corput method."

2.1. Universal Torsor Parametrization

The authors utilize the fact that rational points on a split toric variety $X$ can be lifted to integral points on a universal torsor $X_0$ , which is a quasi-affine toric variety.

Lifting: The counting problem on $X(\mathbb{Q})$ is transformed into counting integral points on $X_0(\mathbb{Z})$ satisfying specific coprimality conditions and height bounds.
Parametrization: The geometry of the fan defining $X$ allows for an explicit parametrization of points on $X_0$ via coordinates in $\mathbb{Z}^n$ . The height function on $X$ lifts to a product of monomials on $X_0$ .

2.2. Effective Equidistribution Conditions

The paper formulates and verifies an Effective Equidistribution (EE) condition (Condition EE) and its counterpart on universal torsors (Condition EEUT).

These conditions assert that the number of integral points in an adelic neighborhood (defined by real and congruence conditions) approximates the Tamagawa measure of that neighborhood with an error term that is uniform with respect to the size of the congruence modulus.
Proposition 4.6 demonstrates that if the torsor satisfies (EEUT), the base variety satisfies (EE).

2.3. The Toric Van der Corput Method

A key technical innovation is the Toric van der Corput method (Theorem 7.1).

Goal: To show that the contribution of integral points lying in "lopsided" boxes (where coordinates are small relative to the height bound) is negligible.
Technique: Instead of standard exponential sum estimates, the authors compare the counting of lattice points to integrals over the Néron–Severi torus $T_{NS}$ . By exploiting the specific toric geometry and the structure of the defining fan, they control the error terms when restricting to regions where certain coordinates are small (e.g., $< (\log B)^A$ ). This allows for precise handling of boundary effects in the counting problem.

2.4. Sieving Techniques

Selberg Sieve: Used to handle "infinitely many" local conditions (e.g., points avoiding a divisor modulo $p$ for all $p$ ). The authors develop an effective version (Theorem 3.11) that captures truncated local conditions modulo prime powers.
Geometric Sieve: An effective version of Ekedahl's sieve is proven (Theorem 1.5), showing that points reducing to a codimension $\ge 2$ subvariety modulo a large prime $p$ are sparse. This relies on a quantitative version of the Lang–Weil estimate for fibers of projections.

3. Key Results

3.1. Main Theorems on Equidistribution

Theorem 1.4 (Effective Equidistribution): For a smooth proper split toric variety $X$ over $\mathbb{Q}$ with globally generated $K_X^{-1}$ , the number of rational points in a standard adelic set $F$ with height $\le B$ is:
$\# \{P \in F \cap T_O(\mathbb{Q}) : H_{tor}(P) \le B\} = \alpha(X)\omega_{tor}(F) B(\log B)^{r-1} + O_{F_\infty, \epsilon} \left( L(F_f)^{r+2\dim X + \epsilon} B(\log B)^{\frac{r-3}{2} + \epsilon} \right)$
Here, $L(F_f)$ is the covering exponent of the finite part of the adelic set. This confirms Principle 1.1 with an explicit error term depending polynomially on the size of the adelic neighborhood.
Theorem 1.5 (Geometric Sieve): For a Zariski closed subset $Z \subset X$ of codimension $\ge 2$ , the number of points reducing to $Z$ modulo some prime $p \ge N$ is bounded by:
$\ll_Z \frac{B(\log B)^{r-1}}{N \log N} + B(\log B)^{r-2} \log \log B$
This establishes the "purity" of the distribution: points do not concentrate on codimension $\ge 2$ subvarieties.

3.2. Applications to Fibrations and Local Conditions

Theorem 1.6 (Fibrations): The paper provides quantitative estimates for rational points in the image of a fibration $f: Y \to X$ $f : Y \to X$ .
- If $f$ is generically finite of degree $>1$ , the count is $O(B(\log B)^{r-1-\iota})$ .
- If $f$ has geometrically integral generic fibers, the count depends on the splitting behavior of fibers over codimension 1 points. If all fibers are pseudo-split, the points are equidistributed with a positive density; if non-pseudo-split fibers exist, the density is zero (with a specific logarithmic decay rate).
Theorem 1.8 (Integral Points and Primes): The paper estimates the number of points where a section $s$ takes values in $S$ -integers or prime values. It shows that the density of points where a coordinate takes a prime value is negligible ( $O(B(\log B)^{r-1} / (\log \log B)^\theta)$ ).

3.3. Purity of Equidistribution

Principle 1.3: The authors prove that the equidistribution principle holds for any Zariski open dense subset $W \subset X$ whose complement has codimension $\ge 2$ . This is a quantitative version of the Arithmetic Purity of Strong Approximation (APSA).

4. Significance and Claims

The paper claims the following significance:

First Proof for Toric Varieties: It provides the first proof of the Manin–Peyre equidistribution principle for smooth proper split toric varieties, a class of varieties where Manin's conjecture was known but the finer equidistribution property was not.
Effectiveness: Unlike previous non-effective proofs, this work provides asymptotic formulas with effective error terms. Crucially, the error term explicitly depends on the "size" (covering exponent) of the adelic neighborhood, allowing for uniform results.
Geometric Sieve: It establishes the first effective geometric sieve for higher-dimensional varieties of degree $>1$ (beyond quadratic hypersurfaces), showing that rational points do not cluster on codimension $\ge 2$ subvarieties.
Quantitative APSA: It offers a quantitative version of the Arithmetic Purity of Strong Approximation for split toric varieties, linking the distribution of points to the geometry of the variety and the splitting of fibers in fibrations.
Methodological Innovation: The "toric van der Corput method" is presented as a new tool for handling lattice point counting in spike-shaped domains, complementing existing approaches like harmonic analysis and standard sieving.

The authors note that while their method relies on the split condition (unlike some recent work by Santens which handles non-split cases), their approach yields effective error terms with explicit dependencies on adelic neighborhoods, which appears difficult to extract from other methods. They also acknowledge that their results for non-split cases or more general line bundles are not covered, and that establishing non-trivial lower bounds for points with prime values remains an open problem related to the Schinzel hypothesis.