🔢 mathematics

Signs of Square-Free Fourier Coefficients of Half-Integral weight cusp forms and the Congruent Number Problem

This paper establishes that the square-free Fourier coefficients of half-integral weight cusp forms with a nonzero Shimura lift change sign infinitely often, specifically proving that both positive and negative signs occur at least $\gg X^{4/7-\varepsilon}$ times up to $X$ , with a direct application demonstrating this sign distribution for the weight $3/2$ form associated with the congruent number elliptic curve $y^2=x^3-x$ .

Original authors: Wei Tao, Guo Xuejun

Published 2026-07-28

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Wei Tao, Guo Xuejun

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

The Secret Language of Numbers and the Mystery of the "Congruent" Triangle

Imagine the world of mathematics as a vast, infinite library where every book is a number, and the stories inside are written in patterns. Some of these patterns are like the rhythm of a drumbeat, predictable and steady, while others are like the chaotic splash of a wave, seemingly random. In this library, there is a special section dedicated to "modular forms." Think of these as incredibly complex musical instruments that, when played, produce a specific sequence of numbers called "Fourier coefficients." These numbers aren't just random; they hold deep secrets about the shape of space and the behavior of prime numbers.

One of the most famous riddles in this library is the "Congruent Number Problem." It asks a simple question: Can you build a right-angled triangle using only rational numbers (fractions) for the sides, such that the area of that triangle is a specific whole number? For example, is there a right triangle with an area of 5? Or 6? Or 7? Mathematicians have known for centuries that the answer depends on a hidden code. If a number is "congruent," it means a certain mathematical curve (an elliptic curve) associated with it is "alive" and moving. If it's not, the curve is "frozen." The paper we are about to explore dives into the sign of the numbers produced by these musical instruments. Do they dance up and down, or do they just march in one direction? Understanding this helps us figure out which numbers are "congruent" and which are not.

The Paper's Discovery: A Dance of Signs

In this paper, Tao Wei and Xuejun Guo act like detectives investigating the behavior of these musical instruments, specifically a type called "half-integral weight cusp forms." These instruments are tricky because they don't always play the same tune; sometimes they play a positive note, and sometimes a negative one. The authors wanted to know: If you listen to the sequence of numbers these instruments produce, do the positive and negative notes appear randomly, or is there a bias? Do they eventually stop changing and just pick one side?

The researchers focused on a very specific condition: they looked only at the "square-free" numbers in the sequence. A square-free number is one that isn't divisible by any perfect square (like 4, 9, or 16). Think of these as the "pure" notes in the melody, stripped of any repetitive, squared-up background noise. The team proved a powerful mathematical fact: for these specific instruments, the positive and negative notes are both incredibly abundant. They showed that up to any huge number $X$ , the count of positive square-free notes is at least roughly $X^{4/7}$ , and the count of negative ones is also at least roughly $X^{4/7}$ .

To put that in perspective, if you were to list a billion ( $10^9$ ) of these numbers, the paper proves that there are millions of positive ones and millions of negative ones. The most exciting part of their finding is that this means the signs change infinitely often. The melody never settles into a single mood; it keeps flipping between positive and negative forever. This is a "proven" fact in the paper, derived from rigorous mathematical logic, not just a guess or a simulation.

The Congruent Number Connection

The authors then applied this discovery to the famous Congruent Number Problem. They looked at a specific musical instrument (a weight 3/2 cusp form) that is directly linked to the curve $y^2 = x^3 - x$ . This curve is the "parent" of all the curves used to test if a number is congruent. The numbers this instrument produces are related to the difference between two ways of counting how many ways a number can be written as a sum of squares (using two specific formulas, $Q_1$ and $Q_2$ ).

The paper proves that for odd, square-free numbers, the instrument produces both positive and negative results in huge quantities. Specifically, there are at least roughly $X^{4/7}$ odd square-free numbers up to $X$ where the first counting method wins, and the same huge number of cases where the second method wins.

This has a direct consequence for the Congruent Number Problem. The paper explains that if a number is counted in these results (meaning the instrument produced a non-zero result), that number is not a congruent number. In other words, the paper proves that there are infinitely many numbers that are not congruent numbers, and it gives us a way to find them by looking at the sign changes of these coefficients.

What the Paper Doesn't Say (And What It Suggests)

While the paper proves that both signs appear infinitely often and in large numbers, it does not claim to have solved the entire Congruent Number Problem. It doesn't give us a simple rule to say "Yes, 5 is congruent" or "No, 7 is not" for every single number. Instead, it provides a statistical guarantee: the "dance" of signs never stops.

The authors also looked at the data numerically. They calculated the signs for numbers up to one million and found something fascinating. The number of positive results and the number of negative results were almost exactly equal—hovering right around 50% each. Based on this data, they suggest (but do not prove) that as you go to infinity, the two signs will be perfectly balanced, appearing with equal frequency. This is a conjecture, a hopeful guess based on the evidence they gathered, but the hard proof remains that they both happen infinitely often.

In short, Wei and Guo have shown us that the mathematical music behind the Congruent Number Problem is a lively, ever-changing duet between positive and negative forces, never settling into a single note, and ensuring that the mystery of which numbers are congruent remains a rich and complex landscape.

Technical Summary: Signs of Square-Free Fourier Coefficients of Half-Integral Weight Cusp Forms and the Congruent Number Problem

Problem Statement
The paper investigates the sign distribution of square-free Fourier coefficients of half-integral weight cusp forms. Specifically, it addresses the problem of determining whether the number of positive and negative coefficients up to a bound $X$ is sufficiently large to guarantee infinitely many sign changes. This inquiry is motivated by the Congruent Number Problem, where the sign of a specific weight $3/2$ coefficient, $\Delta(t)$ , distinguishes between the inequalities $r_{Q_1}(t) > r_{Q_2}(t)$ and $r_{Q_1}(t) < r_{Q_2}(t)$ for odd square-free integers $t$ . While Waldspurger's formula relates the square of these coefficients to the central $L$ -value $L(E_t, 1)$ , it does not determine their signs. Previous work by Hulse, Kıral, Kuan, and Lim established infinite sign changes for level 4 forms, but extending these results to general levels required overcoming significant analytic hurdles regarding additive twists.

Methodology
The authors employ a combination of analytic number theory techniques, specifically focusing on the continuation of Dirichlet series and the Rankin–Selberg method.

Functional Equations for Additive Twists: A primary technical innovation is the derivation of functional equations for additive twists of half-integral weight forms at arbitrary cusps for a general level $N$ (where $4|N$ ). Unlike previous level 4 analyses that treated cusps case-by-case (based on the denominator of the cusp), the authors establish a unified functional equation (Lemma 2.2) relating the twist at a rational point $u/d$ to the Fourier expansion at a corresponding cusp. This involves constructing specific scaling matrices and utilizing the metaplectic cover to handle the automorphy factors.
Analytic Continuation of Square-Free Series: Using the derived functional equations, the authors continue the Dirichlet series of square-free coefficients, $M_f(s) = \sum_{t \text{ sq-free}} a_f(t)t^{-s}$ , to the half-plane $\Re(s) > 3/4$ . This is achieved by expressing the series via additive twists, applying the functional equations to move the argument to the critical line, and utilizing the Phragmén–Lindelöf principle to bound the growth in the critical strip.
Rankin–Selberg Second Moments: The paper establishes a lower bound for the second moment of square-free coefficients, $B_f(X) = \sum_{t \le X, \text{sq-free}} |a_f(t)|^2$ . By relating the sum over all integers to the sum over square-free integers (using a transfer argument similar to Lau, Royer, and Wu) and applying the spectral theory of Eisenstein series, they prove that $B_f(X) \asymp_f X$ .
Sign Distribution Argument: Combining the upper bound for the signed sum $S_f(X) = \sum a_f(t)W(t/X)$ (derived from the analytic continuation to $\Re(s) > 3/4$ ) with the lower bound for the absolute sum $A_f(X)$ , the authors demonstrate that the positive and negative contributions must both be substantial. This forces the existence of a large number of coefficients of each sign.

Key Contributions and Results

General Level Sign Theorem (Theorem 1.1): The authors prove that for a half-integral weight cusp form $f \in S_{k/2}(\Gamma_0(N))$ with $4|N$ , satisfying eigenform conditions, having a nonzero cuspidal Shimura lift, and (for $k=3$ ) being orthogonal to the unary theta subspace, the number of positive and negative square-free coefficients up to $X$ satisfies:
$P_f^\pm(X) \gg_{f,\varepsilon} X^{4/7-\varepsilon}$
for every $\varepsilon > 0$ . Consequently, the coefficients change sign infinitely often.
Application to the Congruent Number Problem (Theorem 1.2): Applying the general theorem to the specific weight $3/2$ form $\Psi$ associated with the elliptic curve $E: y^2 = x^3 - x$ , the paper proves that for odd square-free integers $t \le X$ :
$\#\{t \le X : r_{Q_1}(t) > r_{Q_2}(t)\} \gg_\varepsilon X^{4/7-\varepsilon}$
and
$\#\{t \le X : r_{Q_1}(t) < r_{Q_2}(t)\} \gg_\varepsilon X^{4/7-\varepsilon}.$
This confirms that $\Delta(t)$ changes sign infinitely often.
Non-Congruent Numbers: The authors note that since $L(E_t, 1) \neq 0$ for these $t$ (implied by $\Delta(t) \neq 0$ ), the Coates–Wiles theorem implies that the Mordell–Weil rank of $E_t$ is zero. Thus, every integer counted in Theorem 1.2 is a non-congruent number.

Significance
The paper's primary significance lies in extending the known results on sign changes of square-free coefficients from level 4 to a general level $N$ divisible by 4. The authors explicitly state that the "main new issue" in this generalization is the treatment of additive twists at arbitrary cusps, which they resolve by establishing new functional equations.

Regarding the Congruent Number Problem, the paper provides a rigorous proof that the arithmetic function distinguishing the two representation counts ( $r_{Q_1}$ and $r_{Q_2}$ ) is not biased toward one sign; both inequalities occur frequently. While the paper does not provide a new arithmetic interpretation for the sign of $\Delta(t)$ itself (noting that such an interpretation is currently unknown), it establishes the density of non-congruent numbers arising from this specific criterion.

The authors remain modest regarding the distribution of signs, presenting numerical data (Table 1) that suggests the signs are asymptotically equidistributed ( $N_+(X) \sim N_-(X)$ ), but they formulate this as Conjecture 5.1 rather than a proven result. The paper concludes that the sign changes are infinite and that the counts of positive and negative coefficients are of the same order of magnitude, $X^{4/7-\varepsilon}$ .

The Secret Language of Numbers and the Mystery of the "Congruent" Triangle

The Paper's Discovery: A Dance of Signs

The Congruent Number Connection

What the Paper Doesn't Say (And What It Suggests)

More like this