🔢 mathematics

On the Hardy--Littlewood--Chowla conjecture

This paper proves the long-standing Hardy–Littlewood–Chowla conjecture asserting that the Möbius function exhibits cancellation on shifted primes, thereby demonstrating that the values $\mu(p+2)$ are asymptotically equidistributed between $1$ and $-1$ among square-free shifted primes.

Original authors: Guangshi Lü, Zhiwei Wang, Jie Wu

Published 2026-10-15

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Guangshi Lü, Zhiwei Wang, Jie Wu

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

In the vast landscape of number theory, where mathematicians study the hidden architecture of whole numbers, two fundamental forces constantly interact: addition and multiplication. Addition is the simple act of counting up, while multiplication is the act of building numbers from smaller parts, specifically prime numbers. For centuries, mathematicians have wondered how these two forces influence one another. Do they behave independently, or do they conspire to create patterns? To explore this, researchers often look at a special tool called the Möbius function. This tool acts like a switch that assigns a value of positive one, negative one, or zero to every whole number, depending on how its prime factors are arranged. If a number has a square factor, the switch is off. If it has an even number of distinct prime factors, it is positive; if odd, it is negative. The central mystery is whether these values cancel each other out when summed up, suggesting that the prime factors are distributed in a way that is essentially random and unpredictable.

A famous conjecture, known as the Hardy–Littlewood–Chowla conjecture, proposes that this randomness holds true even when we shift our focus. Specifically, it asks what happens if we look at prime numbers and add a small, fixed number to them, such as adding two to a prime to see what the Möbius function says about the result. This is a difficult problem because it sits right at the edge of what current mathematical tools can handle, a place where the usual methods of sieving out numbers fail to distinguish between different types of randomness. For decades, this specific case remained unproven, serving as a benchmark for the limits of our understanding of how primes and their neighbors behave.

In a recent paper, a team of mathematicians has finally solved this specific case. They proved that when you take all the prime numbers up to a very large limit, add two to each of them, and sum up the values of the Möbius function for those results, the total is effectively zero. In other words, the positive and negative values cancel each other out almost perfectly. This confirms that the Möbius function exhibits no hidden bias when applied to primes shifted by two. The researchers did not just show that the sum is small; they demonstrated that it is vanishingly small compared to the total number of primes, proving that the behavior is indeed random as the numbers get larger.

To reach this conclusion, the authors had to navigate a complex mathematical terrain. They broke the problem down into smaller, manageable pieces, using a technique that separates the numbers into different categories based on their size and structure. They introduced a set of auxiliary weights, which can be thought of as a way to smooth out the data and highlight the underlying patterns without getting lost in the noise of individual numbers. By carefully analyzing how these weighted sums interact, they were able to show that the contributions from different parts of the number line cancel each other out with extreme precision. Their method involved a sophisticated interplay between the distribution of prime factors and the properties of the Möbius function, requiring them to prove that certain complex sums, which had previously been intractable, are actually negligible.

The implications of this proof extend beyond just the sum itself. The researchers also investigated the distribution of the positive and negative values among the primes. They found that among the primes for which the shifted number is not zero, the values of positive one and negative one appear with equal frequency. Specifically, as you look at larger and larger sets of primes, exactly half of the non-zero results will be positive, and the other half will be negative. This confirms a long-held intuition that there is no preference for one sign over the other in this context. The constant that governs the density of these square-free shifted primes was also calculated, providing a precise measure of how often these numbers appear.

This work represents a significant step forward in understanding the deep connections between addition and multiplication. By proving that the Möbius function behaves randomly on shifted primes, the authors have removed a major obstacle in the field. Their result suggests that the primes do not conspire to create hidden patterns when shifted, reinforcing the idea that the distribution of prime numbers is fundamentally chaotic and independent of simple arithmetic shifts. The proof relies on rigorous logic and careful estimation rather than simulation or guesswork, offering a definitive answer to a question that has stood for decades. It stands as a testament to the power of modern analytic number theory to resolve questions that were once thought to be beyond reach, bringing clarity to the intricate dance of numbers that underlies our mathematical universe.

Technical Summary: On the Hardy–Littlewood–Chowla Conjecture

Problem Statement
The paper addresses the Hardy–Littlewood–Chowla conjecture, a hybrid problem combining the additive structure of primes with the multiplicative structure of the Möbius function $\mu(n)$ . Specifically, the authors prove the case where the Möbius function is evaluated at shifted primes. The conjecture asserts that for distinct integers $h_1, \dots, h_k$ and $a_1, \dots, a_\ell$ , the sum
$\sum_{n \le x} \mu(n+h_1)\cdots\mu(n+h_k)\Lambda(n+a_1)\cdots\Lambda(n+a_\ell) = (S_{k,\ell} + o(1))x$
where $S_{k,\ell} = 0$ if $k > 0$ . The paper focuses on the specific instance $k=1, \ell=1, a_1=0, h_1=2$ , which corresponds to the cancellation of the Möbius function on shifted primes:
$\sum_{p \le x} \mu(p+2) = o(\pi(x)).$
This problem is noted as a "folklore conjecture" and a well-known model case for the parity problem in sieve theory. Previous work by Tao and Teräväinen (2022) established the asymptotic under the assumption of Siegel zeros, while Lichtman and Teräväinen provided averaged results. This paper provides an unconditional proof for the specific shift $h_1=2$ .

Methodology
The proof employs a sophisticated combination of analytic number theory techniques, primarily centered on the Vaughan identity decomposition of the von Mangoldt function $\Lambda(n)$ and the use of auxiliary weights to handle the parity problem.

Vaughan Decomposition: The authors decompose the sum involving $\Lambda(n)$ into Type I and Type II sums (Proposition 2.1).
- Type I Sums: These involve sums where one variable is small. The authors prove that these sums, when weighted by auxiliary functions $W(n)$ , are negligible (Proposition 3.1). This relies on a Bombieri–Vinogradov type theorem for the Möbius function and careful truncation of divisor sums.
- Type II Sums: These are bilinear forms involving two large variables. This is the core difficulty. The authors expand the auxiliary weights $W(n)$ (defined via products of primes in specific dyadic intervals) and apply Cauchy's inequality.
Auxiliary Weights and State Space: To overcome the parity obstruction, the authors introduce auxiliary weights $W(n)$ constructed from disjoint sets of primes $P_i$ in intervals $(\exp(L^{a_i}), \exp(2L^{a_i}))$ . These weights are designed to have specific distribution properties modulo primes.
Operator-Theoretic Approach (The Type II Estimate):
- The Type II sum is reduced to estimating a quadratic form involving a determinant condition $bmn - ars = 2(a-b)$.
- The authors encode this arithmetic condition into a state space of integer vectors and prime lists. They define operators ( $G_j, E_j$ ) acting on this space, representing transitions between states (adding/removing prime factors).
- Parity Conjugacy: A crucial step (Proposition 4.3) involves a linear map $T$ that transforms the problem from a "dirty" condition involving $-2(b-a)$ to a "clean" condition involving $b-a$ . This allows the use of major arc analysis on a transformed domain.
- Root Averaging: The authors average over the "roots" (parameters defining the state space) and replace arithmetic conditions with integrals over finite fields (Proposition 4.4). This reduces the problem to bounding moments of auxiliary operators.
- Moment Bounds: By constructing a sequence of operators and proving that "dirty" edges (where prime lists overlap in specific ways) have small norms, they show that the total moment is bounded by $L^{-2R E_0}$ (Proposition 4.5). This relies on the distinctness of prime selections and the rapid decay of operator norms.
Major Arc Analysis: The major arc contribution is bounded using character–Mellin estimates (Proposition 2.2) and large-value estimates for polynomials over primes (Lemma 5.2). The authors show that the major arc contribution is also negligible (Proposition 5.3).

Key Contributions and Results

Theorem 1: The authors prove unconditionally that
$\sum_{p \le x} \mu(p+2) = o(\pi(x)).$
This confirms the conjecture for the shift $h_1=2$ .
Theorem 2 (Distribution of Signs): As a direct application, the paper establishes the asymptotic density of the signs of $\mu(p+2)$ . Let $A_2 = \prod_{p>2} (1 - \frac{1}{p(p-1)})$ . Then for $\kappa \in \{-1, 1\}$ :
$\#\{p \le x : \mu(p+2) = \kappa\} = \left(\frac{A_2}{2} + o(1)\right)\pi(x).$
Consequently, among the primes $p$ for which $\mu(p+2) \neq 0$ (i.e., $p+2$ is square-free), the values $1$ and $-1$ occur with asymptotic density $1/2$ each:
$\lim_{x \to \infty} \frac{\#\{p \le x : \mu(p+2) = \kappa\}}{\#\{p \le x : \mu^2(p+2) = 1\}} = \frac{1}{2}.$

Significance
The paper claims to resolve a long-standing open problem in analytic number theory, specifically the case of the Hardy–Littlewood–Chowla conjecture for shifted primes with shift 2. The significance lies in:

Unconditional Proof: Unlike previous partial results that relied on the existence of Siegel zeros, this proof is unconditional.
Parity Problem: The result is described as a well-known model case for the parity problem in sieve theory. The successful application of the auxiliary weight method combined with the operator-theoretic framework demonstrates a powerful new approach to handling multiplicative correlations in additive settings.
Distribution of Signs: The paper provides the first proof that the Möbius function takes positive and negative values with equal frequency on the set of shifted primes (where it is non-zero), addressing a question related to Sarnak's Problem 5.2 (originally posed for the Liouville function).

The authors restrict their main theorem to the shift $h_1=2$ for simplicity and its connection to the twin prime conjecture, noting that the method is tailored to this specific case. The paper does not claim to generalize the result to arbitrary shifts or higher $k, \ell$ without further development.

More like this