🔢 mathematics

Counting Salem numbers arising from arithmetic hyperbolic orbifolds

This paper advances the quantitative study of arithmetic hyperbolic orbifolds by establishing bounds on the proportion of Salem numbers within specific commensurability classes and improving lower bounds for the growth of geodesic length multiplicities, achieved through new estimates on Salem numbers with fixed discriminant properties and a generalized counting method for Pythagorean triples.

Original authors: Michelle Chu, Plinio G. P. Murillo, Otto Romero, Lola Thompson

Published 2026-03-27

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Michelle Chu, Plinio G. P. Murillo, Otto Romero, Lola Thompson

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 an explorer trying to map a vast, invisible ocean. This ocean isn't made of water, but of numbers and shapes that exist in higher dimensions. Specifically, you are looking for a very special type of number called a Salem number.

Think of a Salem number as a "golden key." It's a number slightly bigger than 1, but its "siblings" (mathematical cousins) are all stuck on a perfect circle, neither growing nor shrinking. These keys are rare and mysterious, and mathematicians have been trying to figure out how many of them exist and where they hide.

This paper is like a new, high-tech sonar system designed to count these keys, but with a specific twist: it only counts the keys that fit into a very specific type of lock found in arithmetic hyperbolic orbifolds.

Here is a breakdown of what the authors did, using simple analogies:

1. The Lock and the Key (The Connection)

In the world of geometry, there are shapes called hyperbolic orbifolds. You can think of these as complex, multi-dimensional surfaces (like a hyperbolic saddle shape that goes on forever).

The Key: A Salem number represents the length of a "closed loop" (a geodesic) you can draw on this surface. If you walk along this loop, you return to your start, and the number tells you how long the trip was.
The Lock: These loops only exist on surfaces built from very specific, rigid mathematical rules (called arithmetic lattices).
The Problem: Not every Salem number can be a key for every lock. The authors wanted to know: If we look at all the locks in a specific dimension (say, 3D, 5D, 7D), what percentage of the available keys actually fit?

2. The "Pythagorean Triples" Detour

To count the keys, the authors had to solve a different, classic puzzle first. They needed to count how many ways you can solve an equation that looks like the famous Pythagorean theorem ( $A^2 + B^2 = C^2$ ), but with a twist: $A^2 + D \times B^2 = C^2$ .

Think of this as counting Pythagorean Triples with a weird weight.

In the standard version, you are looking for integer solutions (whole numbers) that fit a triangle.
In this paper, the "weight" $D$ changes the shape of the triangle.
The authors developed a new, elementary way to count these solutions. They used a method similar to Gauss's lattice counting, which is like counting how many dots (integer coordinates) fit inside a growing circle or ellipse. They proved that even with this weird weight, you can predict exactly how many solutions exist as the numbers get bigger.

3. The Main Discovery: The "Filter"

Once they could count the solutions to that equation, they applied it to the Salem numbers.

The Old View: Before this, we knew there were lots of Salem numbers, but we didn't know how many of them were "real" (i.e., actually corresponded to a physical loop on these geometric shapes).
The New View: The authors proved that the number of "real" Salem numbers is actually quite small compared to the total number of possible Salem numbers.
The Analogy: Imagine a giant jar filled with millions of marbles (all possible Salem numbers). Most are just glass beads. The authors found a sieve (the condition $f(1)f(-1) = -D$ ) that only lets the "golden" marbles (the ones that fit the geometric locks) fall through. They calculated exactly how many golden marbles pass through the sieve as the jar gets bigger.

4. Why Does This Matter? (The "Traffic Jam" of Loops)

The paper also looks at multiplicities.

Imagine a highway (the geometric shape). Usually, you expect one car (one loop) per lane.
But in these arithmetic shapes, sometimes many different loops have the exact same length. It's like a traffic jam where 100 cars are all stuck at the exact same mile marker.
The authors showed that as the highway gets longer (as the length of the loops increases), the number of cars stuck at the same spot grows exponentially.
The Result: They proved that for these specific shapes, the "traffic jam" gets incredibly dense very quickly. The average number of loops sharing the same length grows at a rate of roughly $e^{(n-1)\ell/2}$ . This is a massive explosion in complexity.

Summary of the "Big Picture"

The Goal: Count how many special numbers (Salem numbers) actually correspond to real geometric loops in specific high-dimensional shapes.
The Method: They translated the problem into counting solutions to a weighted version of the Pythagorean theorem. They used a clever "dot-counting" technique to solve this.
The Result: They found that while there are infinite Salem numbers, the ones that fit these specific geometric locks are rare and follow a very precise, predictable pattern.
The Bonus: They proved that on these shapes, you get "traffic jams" of loops—many different paths having the exact same length—and these jams get exponentially worse as the paths get longer.

In short, the authors built a better map for the "ocean of numbers," showing us exactly where the "golden keys" are hidden and revealing that the paths they unlock are surprisingly crowded.

1. Problem Statement and Context

The paper investigates the quantitative relationship between Salem numbers and the geometry of arithmetic hyperbolic orbifolds.

Background: Salem numbers are real algebraic integers $\lambda > 1$ whose Galois conjugates lie on the unit circle (except $\lambda^{-1}$ ). They are deeply connected to the lengths of closed geodesics in arithmetic hyperbolic manifolds. Specifically, if $\gamma$ is a hyperbolic isometry in a classical arithmetic lattice $\Gamma \subset \text{Isom}(\mathbb{H}^n)$ , then $\lambda = e^{\ell(\gamma)}$ is a Salem number.
The Gap: While the existence of this link is known, the proportion of Salem numbers that actually arise from specific arithmetic lattices (specifically those defined over $\mathbb{Q}$ in odd dimensions) was not well-quantified. Furthermore, the growth rate of the multiplicity average of the geodesic length spectrum (how many geodesics share the same length) needed sharper bounds for non-compact orbifolds.
Core Question: What is the proportion of Salem numbers of degree up to $n+1$ that are realized by classical arithmetic lattices in odd dimension $n$ ?

2. Methodology

The authors employ a multi-step approach combining algebraic number theory, the geometry of quadratic forms, and analytic number theory (lattice point counting).

A. Algebraic Translation to Diophantine Equations

The central technical innovation is translating the condition for a Salem polynomial $f(x)$ to arise from an arithmetic lattice into a Diophantine counting problem.

The Condition: Based on work by Bayer-Fluckiger, a necessary condition for a polynomial $f$ to be the characteristic polynomial of an isometry in a quadratic space with reduced determinant $-D$ is:
$f(1)f(-1) \equiv -D \pmod{\mathbb{Q}^{\times 2}}$
The Reduction: Since $f$ is a Salem polynomial with integer coefficients, $f(1)f(-1)$ is an integer. The condition implies $f(1)f(-1) = -D k^2$ for some integer $k$ .
Variable Definition: Let $A = \sum_{k \text{ even}} c_k$ and $C = \sum_{k \text{ odd}} c_k$ . Then $f(1)f(-1) = A^2 - C^2$ . The condition becomes:
$A^2 - C^2 = -D k^2 \implies A^2 + D k^2 = C^2$
This transforms the problem into counting integer solutions to the generalized Pythagorean equation $A^2 + DB^2 = C^2$ (where $B=k$ ).

B. Counting Diophantine Solutions (Theorem 3.1)

The authors derive an asymptotic formula for the number of primitive integer solutions to $A^2 + DB^2 = C^2$ with $|C| \le X$ .

Technique: They use a geometric parameterization of the ellipse $A^2 + DB^2 = 1$ combined with Gauss's lattice point counting argument.
Novelty: They provide an elementary proof generalizing the count of Pythagorean triples. They distinguish cases based on the parity of $D$ and the parity of the solutions, utilizing the Möbius function to isolate primitive solutions.
Result: They establish that the number of solutions grows linearly with $X$ , with a constant factor depending on the number of prime factors of $D$ and $\sqrt{D}$ .

C. Bounding Polynomial Coefficients

To count the Salem polynomials themselves, the authors bound the number of coefficient tuples $(c_0, \dots, c_{2m})$ that satisfy the Diophantine condition and the Mahler measure bound ( $M(f) \le Q$ ).

They utilize Lemma 4.3 (a result on partition functions) to count the number of integer solutions to linear equations derived from the coefficient sums $A$ and $C$ .
They separate the count into irreducible (Salem) and reducible polynomials, showing the reducible ones are negligible in the asymptotic limit.

D. Application to Geodesics

Finally, they map the count of Salem numbers back to the length spectrum of the orbifold using the relation $\ell = \log \lambda$ .

3. Key Contributions and Results

Theorem A: Counting Salem Polynomials

The authors bound the number of Salem polynomials $f$ of degree $2m$ with Mahler measure $\le Q$ satisfying $f(1)f(-1) \equiv -D$ .
$\# \{f\} \le \frac{\kappa(m, D)}{\pi \sqrt{D}} Q^{m-1} \log Q + O(Q^{m-1})$
This establishes that the density of such Salem numbers is significantly lower than the total set of Salem numbers of that degree.

Theorem B: Proportion in Commensurability Classes

For a classical arithmetic lattice in odd dimension $n$ (where $2m = n+1$ ) defined over $\mathbb{Q}$ with reduced determinant $-D$ , the number of realized Salem numbers $\lambda \le Q$ is bounded by the same asymptotic formula as Theorem A.

Significance: This quantifies the "rarity" of Salem numbers arising from specific arithmetic lattices. It implies that as the degree increases, the proportion of Salem numbers realizable by a fixed commensurability class tends to zero.

Corollaries C & D: Bounds on Geodesic Lengths

These corollaries translate the polynomial counts into bounds on the number of distinct geodesic lengths $\le L$ .

For cocompact (or classical) lattices: The number of distinct lengths grows as $O(e^{(m-1)L} L)$ .
For non-classical lattices (using the subgroup $\Gamma(2)$ ): The growth is $O(e^{2(m-1)L} L)$ .
This provides explicit upper bounds on the complexity of the length spectrum for these manifolds.

Theorem E: Lower Bound on Multiplicity Averages

The paper proves a lower bound for the average multiplicity $G(\ell)$ of the length spectrum for non-compact arithmetic orbifolds of odd dimension $n > 4$ .
$G(\ell) \ge c \cdot \frac{e^{\frac{(n-1)\ell}{2}}}{\ell^2}$

Context: Previous work (e.g., Luo-Sarnak for surfaces) established exponential growth for surfaces. This extends the result to higher odd dimensions, showing that the average multiplicity grows exponentially, specifically at a rate related to the dimension.

4. Significance and Impact

Quantitative Geometry: The paper moves beyond qualitative connections between Salem numbers and hyperbolic geometry to provide precise quantitative bounds. It answers "how many" Salem numbers arise from specific arithmetic structures.
Diophantine Generalization: The elementary proof for counting solutions to $A^2 + DB^2 = C^2$ (Theorem 3.1) is a standalone contribution of independent interest to number theory, generalizing the classical Pythagorean triple count.
Lehmer's Conjecture Connection: By reformulating the problem in terms of geodesic lengths, the work contributes to the broader context of Lehmer's conjecture (which posits a lower bound on Mahler measure). The "Short Geodesic Conjecture" (Conjecture 2.4) is framed as the geometric equivalent of Lehmer's conjecture.
Spectral Geometry: The results on multiplicity averages (Theorem E) deepen the understanding of the spectral properties of arithmetic orbifolds, linking the distribution of lengths to the underlying algebraic structure of the lattice.

In summary, this paper successfully bridges the gap between the distribution of algebraic integers (Salem numbers) and the geometric invariants of hyperbolic orbifolds, utilizing advanced lattice counting techniques to derive sharp asymptotic bounds.