Shanks bias in function fields
Seewoo Lee · 2025
Read Shanks bias in function fields by Seewoo Lee (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2509-16142.txt.
Introduction
Number races often hide a bias, where one side of a count stays ahead more than chance would suggest. Seewoo Lee studies such a bias for polynomials over finite fields, in a form tied to the Liouville function and a quadratic character. The race compares how often a signed product lands on the plus side versus the minus side.
Under a broad hypothesis about L-functions, Lee proves the count leans toward the plus side. A few cases where that hypothesis fails are also given. The result extends a classic bias first seen for prime numbers into the world of function fields, with clear limits.
Catalog record
| CREATOR | Seewoo Lee |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 62 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2509-16142
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Shanks bias in function fields Seewoo Lee Abstract We study the function field analogue of Shanks bias. For Liouville function λ ( f ) \lambda(f) , we compare the number of monic polynomials f f with λ ( f ) χ m ( f ) = 1 \lambda(f)\chi_{m}(f)=1 and λ ( f ) χ m ( f ) = − 1 \lambda(f)\chi_{m}(f)=-1 for a nontrivial quadratic character χ m \chi_{m} modulo a monic square-free polynomial m m over a finite field. Under Grand Simplicity Hypothesis (GSH) for L L -functions, we prove that λ ⋅ χ m \lambda\cdot\chi_{m} is biased towards + 1 +1 . We also give some examples where GSH is violated. 1 Introduction Chebyshev’s bias [ 8 ] is a phenomenon in analytic number theory, referring to the observed preference for certain residue classes of primes in arithmetic progressions. Specifically, it refers to the observed tendency for there to be more primes of the form 4n+3 than of the form 4n+1 among the primes up to some bound. For general moduli, it is conjectured that non-quadratic residues are favored over quadratic residues. Rubinstein and Sarnak were the first to prove this under the Generalized Riemann Hypothesis (GRH) and the Grand Simplicity Hypothesis (GSH) for Dirichlet L -functions. Shanks [ 15 ] approached the bias from a different angle, examining higher-order effects. Based on numerical experiments, he conjectured that there are more positive integers n such that \lambda(n)\chi_{-4}(n)=+1 than those with \lambda(n)\chi_{-4}(n)=-1 , where \lambda is the Liouville function and \chi_{-4} is the nontrivial quadratic character modulo 4 (See Section 2.1 for details). Cha [ 4 ] studied a function field analogue of Chebyshev’s bias, closely following Rubinstein–Sarnak’s argument [ 14 ] , and proved the existence of bias under similar assumptions for L -functions. He also exhibited examples with biases in unexpected directions when L -functions do not satisfy GSH. Later works have extended these ideas to Galois extension of function fields [ 6 ] , explicity formula for the limiting distributions of biases [ 7 ] , multiple prime factors [ 9 ] , and elliptic curves [ 5 ] . In this paper, we study the function field analogue of Shanks conjecture. Let p>2 be a prime and A=\mathbb{F}_{q}[T] be the polynomial ring over a finite field \mathbb{F}_{q} of size q=p^{k} . Let m\in A be a monic square-free polynomial and \chi_{m} be the unique nontrivial quadratic character modulo m . Writing the Liouville function on A as \lambda:A\to\{\pm 1\} , we prove that \lambda\cdot\chi_{m} is biased towards +1 under GSH. This gives an affirmative answer to a function field analogue of Problem 22 in the Comparative Prime Number Theory Problem List [ 10 ] . Theorem (Corollary 4.5 and 4.7 ) . Let A_{\pm}^{\lambda}(n;m) be the number of non-constant monic polynomials f\in A of degree at most n with \lambda(f)\chi_{m}(f)=\pm 1 . Assume that GSH holds for the Dirichlet L -function L(s,\chi_{m}) . Then we have \lim_{n\to\infty}\frac{\#\{1\leq k\leq n:A_{+}^{\lambda}(k;m)>A_{-}^{\lambda}(k;m)\}}{n}>\frac{1}{2} where a closed formula for the density is obtained. When q\equiv 1\pmod{4} , the density can be made arbitrarily close to \frac{1}{2} by choosing m appropriately with large enough degree. We also study several variants (e.g. non-cumulative densities and replacing \lambda by the Möbius function \mu ). In particular, we show there is no bias when restricting to square-free polynomials (Theorem 4.3 ). Example calculations of the densities are given in Section 5 . To prove our main theorem, we first exhibit L -functions attached to these biases (Proposition 3.1 ). We then apply the Kronecker–Weyl equidistribution theorem and Portmanteau’s theorem to derive explicit density formulas via elementary trigonometric-series calculations. In particular, one expresses the bias as a linear combination of cosines with \mathbb{Q} -linearly independent periods; equidistribution reduces the question to evaluating a constant term, and the final inequality follows from comparing symmetric polynomials in those cosines (Theorem 4.4 ). We also gave several numerical examples to illustrate our results (Section 5 ). Supplementary SageMath codes can be found in the GitHub repository https://github.com/seewoo5/sage-function-field . 2 Preliminaries 2.1 Chebyshev’s bias and its variants Since Chebyshev first observed this bias in 1853 [ 8 ] , it has been widely studied by several authors. Littlewood [ 12 ] proved that \pi_{+}(n)>\pi_{-}(n) happens infinitely often, where \pi_{\pm}(n) is the number of primes p\leq n such that p\equiv\pm 1\pmod{4} . Rubinstein and Sarnak [ 14 ] gave more quantitative results. Assuming Generalized Riemann Hypothesis (GRH) and Grand Simplicity Hypothesis (GSH), they proved that the logarithmic density of the set of n such that \pi_{-}(n)>\pi_{+}(n) is about 0.9959 , which is very close to 1 but not exactly. In fact, they considered general “prime race” scenarios with general modulus where more than two congruence classes may exist, and computed the limiting distribution of the normalized bias. Shanks [ 15 ] studied various aspects of Chebyshev’s bias. He considered the normalized bias \tau(n)=\frac{\pi_{-}(n)-\pi_{+}(n)}{\pi(n)}\sqrt{n} and conjectured that \lim_{N\to\infty}\frac{1}{N-1}\sum_{n=1}^{N}\tau(n)=1 supported by numerical evidence. In particular, he considered the higher-order effects and conjectured that, if \pi_{\pm}^{(a)}(n) counts the number of positive integers of the form 4k\pm 1 that are product of a primes (not necessarily distinct), then the average of the order a normalized bias \tau^{(a)}(n)=\frac{\pi_{-}^{(a)}(n)-\pi_{+}^{(a)}(n)}{\pi_{-}^{(a)}(n)+\pi_{+}^{(a)}(n)}\sqrt{n} is (-1)^{a} [ 15 , Section 9] . In other words, there are more numbers of the form 4k-1 (resp. 4k+1 ) with odd (resp. even) number of prime factors than with even (resp. odd) number of prime factors. Specifically, he claimed that the number of n with \lambda(n)\chi_{-4}(n)=+1 exceeds that of \lambda(n)\chi_{-4}(n)=-1 , where \lambda is the Liouville function \lambda(n)=(-1)^{e_{1}+e_{2}+\cdots+e_{k}}\quad\text{when}\quad n=p_{1}^{e_{1}}p_{2}^{e_{2}}\cdots p_{k}^{e_{k}} and \chi_{-4}=\left(\frac{\cdot}{-4}\right) is the nontrivial quadratic Dirichlet character modulo 4 . Note that the L -function for \lambda(n)\chi_{-4}(n) is simply given by \sum_{n\geq 1}\frac{\lambda(n)\chi_{-4}(n)}{n^{s}}=\frac{(1-2^{-2s})\zeta(s)}{L(s,\chi_{-4})} (1) and the bias might be related to the analytic properties of this L -function ( 1 ). The Problem 22 in the comparative prime number theory problem list [ 10 ] asks to compute logarithmic density of the corresponding set (or replacing \chi_{-4} with other quadratic Dirichlet characters), assuming GRH and GSH for the relevant L -functions. 2.2 Function fields, L -functions, and Chebyshev’s bias We fix an odd prime p . Let A=\mathbb{F}_{q}[T] be the polynomial ring over a finite field \mathbb{F}_{q} of size q=p^{k} . Let A^{\circ}\subset A be the subset of monic polynomials. For a monic square-free polynomial m\in A^{\circ} , let \chi_{m}(f):=\left(\frac{f}{m}\right) be the unique nontrivial quadratic character modulo m . Möbius function \mu and Liouville function \lambda are defined on A as \displaystyle\mu(f) \displaystyle:=\begin{cases}(-1)^{k}&f=g_{1}\cdots g_{k},\text{ $f$ is square-free and $g_{1},\dots,g_{k}$ are distinct irreducible factors}\\ 0&f\text{ is not square-free}\end{cases} (2) \displaystyle\lambda(f) \displaystyle:=(-1)^{e_{1}+\cdots+e_{k}},\,\text{if }f\text{ factors as }g_{1}^{e_{1}}\cdots g_{k}^{e_{k}}. (3) Especially, they agree on square-free polynomials. Riemann zeta function for A is defined as \zeta(s)=\sum_{f\in A^{\circ}}\frac{1}{|f|^{s}}=\frac{1}{1-q^{1-s}} where |f|:=q^{\deg f} for f\in A . If \chi is a Dirichlet character on A with modulus m\in A , then the Dirichlet L -function L(s,\chi) is similarly defined as L(s,\chi)=\sum_{f\in A^{\circ}}\frac{\chi(f)}{|f|^{s}}. We often write L -functions as a function in u=q^{-s} and write as \mathcal{L}(u,\chi) . Both functions admit Euler product over monic irreducible polynomials as \displaystyle\zeta(s) \displaystyle=\prod_{P}(1-|P|^{-s})^{-1}, \displaystyle L(s,\chi) \displaystyle=\prod_{P}(1-\chi(P)|P|^{-s})^{-1}. When \chi is a nontrivial character, it is known that the