🔢 mathematics

On the role of higher roots in prime ideal races

This paper introduces a new criterion for Chebyshev's bias in prime ideal races that arises solely from differences in the number of $2p$ -th roots for odd primes $p$ , demonstrating that such bias can occur without relying on square root differences or the order of vanishing of Artin $L$ -functions, while also establishing unconditional results for certain Deep Riemann Hypothesis estimates.

Original authors: Mounir Hayani

Published 2026-07-28

📖 6 min read🧠 Deep dive

CC BY 4.0

Original authors: Mounir Hayani

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 a universe where numbers aren't just cold, hard facts, but characters in a grand, chaotic race. In the world of mathematics, specifically a branch called number theory, there's a famous game called the "Prime Ideal Race." Think of prime numbers as runners on a track. For over a century, mathematicians have noticed that some runners seem to take the lead more often than others, even though, in the long run, they should be perfectly tied. This phenomenon is known as "Chebyshev's bias." It's like flipping a coin and noticing that, for a long time, it lands on heads more often than tails, even though you know it's a fair coin. The big question has always been: Why does this happen? Is it just a fluke, or is there a hidden rulebook governing the race?

For a long time, the answer seemed to be a mix of two things: how many "square roots" a number has (like how many ways you can multiply a number by itself to get a result) and how the race behaves at a specific, tricky point in time called $s = 1/2$ . But what if there's a third, hidden factor? What if the bias comes from something deeper, something involving "higher roots" that no one had noticed before? This is the mystery that Mounir Hayani's paper sets out to solve. The author isn't just guessing; they are building a mathematical machine to prove that this hidden factor exists and to find the smallest possible "engine" (a specific type of group structure) that can make it work.

The Story of the Hidden Root

The paper starts by introducing a new way to look at the race. Imagine the runners (prime numbers) are wearing badges that tell us about their "roots." A "square root" is like a runner who can split into two identical copies of themselves. A "higher root," like a 6th root or a 10th root, is like a runner who can split into six or ten identical copies. For a long time, mathematicians thought the bias in the race was only caused by differences in square roots or by a specific glitch at the halfway point of the race.

Hayani's paper says, "Hold on! We found a new rule." The author introduces two special algebraic parameters (think of them as dials on a control panel) to measure the race. By turning these dials, they show that you can create a race where the bias is entirely caused by a difference in the number of 2p-th roots (where $p$ is an odd prime like 3, 5, or 7). This is a big deal because it proves that the bias can exist even when the old rules (square roots and the halfway glitch) are completely neutral. It's like discovering that a car can speed up not because of the engine or the tires, but because of a secret gear shift no one knew about.

Building the Smallest Possible Machine

The most exciting part of the paper is the construction of the "machines" that make this bias happen. The author asks: "What is the smallest, simplest group of numbers (a Galois group) that can create this specific type of bias?"

To answer this, they had to build a very specific kind of mathematical structure. They used something called "generalized quaternion groups." If you imagine a standard group as a simple circle of dancers, a quaternion group is like a complex, twisting dance where the dancers can flip and spin in ways that break the usual rules of symmetry.

The paper proves that to get this "higher root" bias, you need a machine of a certain minimum size.

For the prime number 3, the smallest machine must have 96 parts.
For the prime number 5, the smallest machine must have 320 parts.

The author didn't just guess these numbers; they proved that no smaller machine could possibly work. They constructed these machines using a direct product of two generalized quaternion groups (imagine two complex dance troupes joining hands to form one massive, intricate troupe). For example, for the prime 3, the machine is made of a group of 12 dancers and a group of 8 dancers, multiplied together to get 96. This is the absolute minimum size required to make the bias appear without any help from square roots or the halfway glitch.

The "Unbiased" Paradox

The paper also tackles a fascinating paradox. There are two ways to define a "fair" race. One way (Rubinstein-Sarnak) looks at the long-term density of the race. The other way (Aoki-Koyama) looks at how the race behaves with a specific weight. The author shows that it is possible to build a race that looks "unbiased" (fair) under the Aoki-Koyama rules but is actually "biased" (unfair) under the Rubinstein-Sarnak rules.

They prove that for certain primes, you can have a race where the difference between the runners stays small and steady (unbiased in one sense), yet one runner is still consistently ahead of the other (biased in the other sense). This happens because the "higher roots" create a subtle, persistent lead that doesn't show up in the other type of measurement. It's like a runner who is always slightly ahead, but the gap never grows large enough to be noticed by a casual observer, only by a mathematician with a very specific ruler.

What's Next?

The paper doesn't just stop at 3 and 5. The author provides a recipe for building these machines for any odd prime. They even offer a formula to guess the smallest size for larger primes, like 7, 11, or 13. For instance, for the prime 11, they predict the smallest machine needs 8,096 parts. While they haven't proven this for every single prime yet, the pattern is strong, and they invite other mathematicians to test their formula.

In short, this paper opens a new door in the world of prime number races. It shows that the bias isn't just about square roots or halfway glitches; it's about a whole new family of "higher roots." By building the smallest possible machines to demonstrate this, the author has given us a clearer, more complete picture of the hidden rules that govern the dance of prime numbers. It's a reminder that in mathematics, even the most familiar races can have secret twists that only the most curious explorers can find.

Technical Summary: On the Role of Higher Roots in Prime Ideal Races

Problem Statement
The paper investigates Chebyshev's bias in the context of prime ideal races within Galois extensions of number fields. Specifically, it addresses the phenomenon where the logarithmic density $\delta_{L/K}(C_1, C_2)$ of the set of $x$ for which the normalized prime ideal counting function $\pi(x; L/K; C_1)$ exceeds $\pi(x; L/K; C_2)$ lies strictly in the interval $(1/2, 1)$ .

Existing literature (e.g., Rubinstein and Sarnak, Ng) has established that such biases typically arise from two primary mechanisms:

A difference in the number of square roots of elements in the conjugacy classes $C_1$ and $C_2$ .
The order of vanishing of Artin $L$ -functions at $s = 1/2$ (specifically, the presence of symplectic representations).

The central problem addressed by this work is whether a bias can be generated solely by the difference in the number of higher-degree roots (specifically $2p$ -th roots for an odd prime $p$ ), in the absence of both square root discrepancies and vanishing $L$ -functions at $s=1/2$ .

Methodology
The paper employs a combination of analytic number theory and explicit group-theoretic constructions.

Analytic Framework:
- The author introduces two algebraic parameters, $\iota(t)$ and $\gamma(t)$ , associated with a class function $t$ on the Galois group $G$ . These parameters measure the smallest squarefree integers $\ell$ for which the induced function of the $\ell$ -th power map is non-zero and for which the inner product with the root-counting function is non-zero, respectively.
- Under standard hypotheses (Generalized Riemann Hypothesis (GRH), Artin's Holomorphicity Conjecture (AC), and Linear Independence (LI) of zeros), the paper derives a conditional characterization (Theorem 1.2) for the bias.
- The analysis utilizes inclusion-exclusion principles relating Chebyshev-type functions to the zeros of Artin $L$ -functions, extending the Rubinstein–Sarnak framework to isolate the contribution of higher roots.
Group-Theoretic Constructions:
- To realize the bias unconditionally, the paper constructs specific Galois extensions where the Galois group $G$ satisfies a property $P(2p; x, y)$ . This property requires that for conjugacy classes $C_1, C_2$ containing $x, y$ , the number of $\ell$ -th roots is equal for all squarefree $\ell < 2p$ , but differs for $\ell = 2p$ .
- The constructions rely heavily on generalized quaternion groups ( $Q_{4n}$ ). The author embeds products of these groups (e.g., $Q_{4p} \times Q_{4(p-1)}$ ) into general linear groups $GL(4, \mathbb{F}_q)$ and subsequently into symmetric groups $S_{q^4}$ to ensure the necessary representation-theoretic constraints (specifically, that the induced class function vanishes for lower powers but not for $2p$ -th powers).
- The paper proves that the Galois group of the extension over $\mathbb{Q}$ must admit no irreducible symplectic representations to ensure the non-vanishing of $L$ -functions at $s=1/2$ .
Lower Bound Proofs:
- The paper establishes rigorous lower bounds on the order of groups satisfying property $P(2p; x, y)$ . This involves analyzing root-counting formulas, utilizing the Schur–Zassenhaus theorem, and applying Wall's classification of groups with a high proportion of involutions.

Key Contributions and Results

Conditional Characterization (Theorem 1.2): The paper provides a necessary and sufficient condition for a prime ideal race to be biased towards $C_1$ ( $\delta > 1/2$ ) in terms of the parameters $\iota(t)$ and $\gamma(t)$ . It isolates the case where the bias is driven entirely by higher roots ( $\gamma(t) = 2\iota(t)$ ) and the vanishing of $L$ -functions is absent.
Unconditional Constructions (Theorem 1.5): The author constructs explicit Galois extensions $L/K$ $L / K$ where the bias arises exclusively from a difference in the number of $2p$ $2 p$ -th roots.
- For $p=3$ , the minimal Galois group order is proven to be 96. The group structure is $Q_{12} \times Q_8$ .
- For $p=5$ , the minimal Galois group order is proven to be 320. The group structure is $Q_{20} \times Q_{16}$ .
- The constructed groups are direct products of generalized quaternion groups and cyclic groups.
Lower Bounds (Theorems 4.1 and 4.2): The paper proves that any finite group satisfying property $P(2p; x, y)$ must have an order of at least $8p^2$ . Specifically, it establishes the optimal lower bounds $|G| \ge 96$ for $p=3$ and $|G| \ge 320$ for $p=5$ , confirming that the constructed examples are of minimal degree.
Weighted Prime Ideal Races (Theorems 1.6 and 1.7): The paper extends the results to weighted prime counting functions $\pi_\alpha(x)$ $π_{α} (x)$ . It demonstrates that for specific extensions, the asymptotic behavior established by Aoki and Koyama under the Deep Riemann Hypothesis (DRH) holds unconditionally under GRH.
- It identifies cases where the race is "unbiased" in the sense of Aoki–Koyama (constant term $K=0$ in the asymptotic expansion) but "biased" in the sense of Rubinstein–Sarnak ( $\delta \in (1/2, 1)$ ), or vice versa, depending on the weight $\alpha$ .
Proposition 1.8: It is shown that for the constructed extensions in $E_p$ , the race is unbiased in the Aoki–Koyama sense for weights $\alpha > 1/(2p)$ , while remaining biased in the Rubinstein–Sarnak sense.

Significance and Claims
The paper claims to resolve a specific gap in the understanding of Chebyshev's bias by demonstrating that the "bias" phenomenon is not limited to square roots or $L$ -function zeros. By constructing examples where the bias is driven purely by the arithmetic of higher roots ( $2p$ -th roots), the author:

Generalizes the Rubinstein–Sarnak criterion: It shows that the bias mechanism is more diverse than previously documented, relying on the interplay between root-counting functions and the induction of class functions.
Provides Minimal Examples: It establishes the absolute minimal group orders required to exhibit this specific type of bias, proving that such phenomena are rare and require complex group structures (non-nilpotent, non-abelian).
Unconditional Results: It removes the need for the Deep Riemann Hypothesis for certain asymptotic estimates regarding weighted prime ideal races, relying only on GRH and the specific group-theoretic constructions.

The paper concludes by posing open questions regarding the general formula for the minimal order $n_p$ for all odd primes and the existence of such groups for all squarefree integers $d$ , suggesting that the generalized quaternion constructions likely yield the minimal structures.

More like this