🔢 mathematics

Asymptotic independence of $\Omega(n)$ and $\Omega(n+1)$ along logarithmic averages

This paper proves the asymptotic independence of the number of prime factors of consecutive integers $\Omega(n)$ and $\Omega(n+1)$ under logarithmic averages with a quantitative double-logarithmic error term, thereby generalizing Tao's result on the Chowla conjecture and yielding new insights into the distribution of $\Omega(p+1)$ for typical almost primes.

Original authors: Dimitrios Charamaras, Florian K. Richter

Published 2026-08-18

📖 6 min read🧠 Deep dive

CC BY 4.0

Original authors: Dimitrios Charamaras, Florian K. Richter

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

Numbers are often thought of as static objects, fixed in their properties, but in the deep study of number theory, they behave more like a vast, shifting landscape. One of the most fundamental ways to describe a number is by counting its prime factors. A prime number is a building block that cannot be divided further, like a single brick. Any whole number can be built by stacking these bricks together. For example, the number 12 is built from three bricks: two 2s and one 3. If we count every brick used to build a number, including duplicates, we get a specific count for that number. Mathematicians call this count the total number of prime factors. For most numbers, this count is surprisingly small, but as numbers get larger, the average number of bricks needed to build them grows very slowly, following a predictable pattern that has been known for decades.

The question that drives this new research is about the relationship between neighbors in this vast landscape. If you pick a number at random and count its bricks, and then look at the very next number and count its bricks, are these two counts related? Do they influence each other, or do they stand alone, completely independent? For a long time, mathematicians have suspected that these two counts are independent, meaning that knowing the count for one number tells you nothing about the count for the next. This idea is a specific version of a much older, famous conjecture about how prime numbers are distributed, which suggests that the signs of certain number patterns cancel each other out completely over time. While this independence has been proven for some specific cases, the general rule for all types of numbers remained unproven, leaving a gap in our understanding of how these fundamental building blocks interact.

In this paper, the researchers have finally filled that gap by proving that these two counts are indeed independent, but they did so with a level of precision that was previously impossible. They focused on a specific way of averaging these numbers, known as a logarithmic average, which gives slightly more weight to smaller numbers and less to larger ones, smoothing out the data to reveal the underlying trend. They showed that for any two functions applied to these counts, the average of their product is equal to the product of their individual averages. In simpler terms, the behavior of the number of prime factors in a number and the number of prime factors in the very next number are statistically unrelated. The researchers did not just prove this is true; they calculated exactly how close the data gets to this independence as the numbers get larger. They found that the error in this approximation shrinks at a very specific, predictable rate, improving upon previous estimates by a factor that, while small, is mathematically significant.

The proof required a sophisticated strategy that combined several different mathematical tools. The researchers first simplified the problem by translating it into the language of multiplicative functions, which are special types of number patterns that behave predictably when numbers are multiplied. They then broke the problem into two parts based on the frequency of the patterns they were analyzing. For patterns that changed slowly, they used a local version of a famous theorem about the distribution of prime factors to show that the independence held. For patterns that changed rapidly, they relied on a powerful, deep result from recent years that deals with how these patterns behave over short intervals. By carefully stitching these two approaches together, they were able to control the error terms and demonstrate that the independence holds true across the board.

One of the most striking aspects of their work is the application to "almost primes," which are numbers that have a specific, small number of prime factors. The researchers showed that if you look at a typical almost prime and then look at the number immediately following it, the number of prime factors in that next number behaves exactly as if you had picked a random number from the entire set of integers. This means that the special structure of being an almost prime does not carry over to its neighbor in a way that affects the count of its prime factors. This finding provides a clearer picture of the randomness inherent in the distribution of prime factors, confirming that the local structure of a number does not dictate the structure of its immediate neighbor.

The paper also addresses a related question about uniform distribution, which asks whether pairs of numbers, when viewed through a specific lens, spread out evenly across a range. The authors proved that if you take two irrational numbers and multiply them by the prime factor counts of consecutive integers, the resulting pairs will spread out evenly across a circle. This happens if and only if both of the original irrational numbers are indeed irrational. If either of them is a rational number, the pairs will cluster in specific spots rather than spreading out. This result connects the independence of prime factor counts to the broader behavior of numbers in geometry and analysis, showing that the statistical independence found in the prime factors has real consequences for how numbers are distributed in space.

The researchers were careful to note that while their error term is very close to the best possible bound known in mathematics, it is not quite the absolute theoretical limit. The limit is set by the inherent variability of the prime factor counts themselves, a fact established by earlier work. Their result falls just short of this limit by a tiny margin, but it is the first time this level of precision has been achieved for this specific type of correlation. They also acknowledged that a different, more direct proof exists that could achieve the absolute best error term, but they chose to present their own method because it offers a different perspective and uses a distinct set of tools. By providing this alternative approach, they have enriched the mathematical toolkit available for solving similar problems in the future.

Ultimately, this work confirms a long-held intuition about the nature of numbers: that the prime factors of one number do not conspire with the prime factors of the next to create a pattern. The distribution of these building blocks is so chaotic and so independent that even when you look at two numbers side by side, they appear to be strangers. This independence is not just a curiosity; it is a fundamental property of the integers that underpins much of modern number theory. By proving this with such rigor and precision, the authors have closed a chapter on a decades-old question and opened the door to new investigations into how these fundamental patterns interact in more complex settings. The result is a clearer, more confident understanding of the invisible architecture that holds the world of numbers together.

Technical Summary: Asymptotic Independence of $\Omega(n)$ and $\Omega(n+1)$ along Logarithmic Averages

Problem Statement
The paper addresses the joint distribution of the number of prime factors, counted with multiplicity, denoted by $\Omega(n)$ , for consecutive integers $n$ and $n+1$ . Specifically, it investigates the asymptotic independence of the values $\Omega(n)$ and $\Omega(n+1)$ when averaged logarithmically. While the Chowla conjecture posits the absence of correlation for the Liouville function $\lambda(n) = (-1)^{\Omega(n)}$ (a specific case of parity), and the Erdős-Kac theorem describes the Gaussian distribution of $\Omega(n)$ individually, the joint behavior for general bounded functions $a(\Omega(n))$ and $b(\Omega(n+1))$ remained an open problem. The authors aim to prove that for any bounded functions $a, b: \mathbb{N} \to \mathbb{C}$ , the logarithmic average of their product factorizes into the product of their individual averages, up to a small error term.

Methodology
The proof strategy follows the framework established by Tao for the logarithmically averaged Chowla conjecture but adapts it to handle functions that are not necessarily multiplicative. The core approach involves several key steps:

Reduction to Multiplicative Functions: The authors first reduce the problem of bounding the correlation of general functions $a(\Omega(n))$ and $b(\Omega(n+1))$ to bounding correlations of specific completely multiplicative functions. This is achieved using a Fourier analytic argument combined with the quantitative Erdős-Kac theorem and a result by Helfgott and Radziwiłł regarding the distribution of $\Omega(n)$ over shifted primes.
Frequency Decomposition: The problem is decomposed into "small frequencies" and "large frequencies" based on the Fourier coefficients of the functions involved.
- Small Frequencies: For frequencies $\xi$ near zero, the associated multiplicative functions are not necessarily strongly aperiodic. The authors establish quantitative bounds for a local Erdős-Kac theorem to control these terms. They utilize Taylor approximations of the exponential function and the Turán-Kubilius inequality to manage the error terms arising from large primes.
- Large Frequencies: For frequencies $\xi$ away from zero, the associated multiplicative functions are shown to be "strongly aperiodic" in the sense of Matomäki, Radziwiłł, and Tao (MRT). This allows the application of deep results from MRT regarding the mean values of multiplicative functions in short intervals. The proof relies on establishing a quantitative lower bound on the distance (in the pretentious sense) between these functions and twisted Dirichlet characters.
Circle Method and Zero-Free Regions: The analysis of large frequencies employs circle method estimates and Vinogradov-Korobov zero-free regions for the Riemann zeta function and Dirichlet L-functions to ensure the necessary distance bounds from twisted characters.

Key Contributions and Results
The primary result is Theorem A, which establishes the asymptotic independence of $\Omega(n)$ and $\Omega(n+1)$ under logarithmic averaging:
$\mathbb{E}_{\log n \in [N]} a(\Omega(n))b(\Omega(n+1)) = \left( \mathbb{E}_{n \in [N]} a(\Omega(n)) \right) \left( \mathbb{E}_{n \in [N]} b(\Omega(n)) \right) + O\left( \frac{(\log \log \log N)^2}{\sqrt{\log \log N}} \right).$
The error term depends only on the supremum norms of $a$ and $b$ .

Generalization: This result generalizes Theorem 1.2 (Tao's proof of the $k=2$ case of the logarithmically averaged Chowla conjecture) by removing the restriction that the functions must be characters or multiplicative. It validates the "two-point correlation" case of a broader conjecture regarding the independence of $\Omega(n), \Omega(n+1), \dots, \Omega(n+k-1)$ .
Quantitative Improvement: The paper provides a quantitative error term with "double-logarithmic savings," approaching the optimal bound known for the Erdős-Kac theorem (which is $O(1/\sqrt{\log \log N})$ ).
Application to Almost Primes: As a direct application, the authors derive Theorem C, which describes the distribution of $\Omega(n+1)$ when $n$ ranges over $\ell$ -almost primes (integers with exactly $\ell$ prime factors). They show that for "typical" values of $\ell$ (within the range predicted by the Erdős-Kac theorem), the sequence $\Omega(n+1)$ behaves asymptotically like $\Omega(n)$ over all integers.
Equivalence: The paper proves the equivalence between the main correlation theorem (Theorem A) and the result on almost primes (Theorem C).
Uniform Distribution: The results imply that the sequence $(\alpha \Omega(n), \beta \Omega(n+1))$ is uniformly distributed modulo 1 with weights $1/n$ if and only if both $\alpha$ and $\beta$ are irrational.

Significance and Claims
The authors claim that this work unifies two distinct number-theoretic perspectives: the distribution of multiplicative parity values (Chowla's conjecture) and the distribution of the total number of prime factors (Erdős-Kac theorem). By proving that $\Omega(n)$ and $\Omega(n+1)$ are asymptotically independent for any bounded functions, the paper suggests a strong form of randomness in the prime factorization of consecutive integers.

The paper acknowledges that while the error term obtained is slightly suboptimal (off by a factor of $(\log \log \log N)^2$ compared to the conjectured optimal $O(1/\sqrt{\log \log N})$ ), it is sufficient to establish the independence. In Appendix A, the authors present an alternative proof communicated by Harald A. Helfgott that utilizes a proposition from Helfgott and Radziwiłł to achieve the optimal error term $O(1/\sqrt{\log \log N})$ . This alternative approach confirms that the result is sharp in the context of the logarithmic averages.

The work does not claim to solve the full Chowla conjecture for $k \ge 3$ or the general Elliott conjecture, but it provides a significant step forward by resolving the two-point case for general functions, thereby extending the scope of known independence results beyond multiplicative characters.

Technical Summary: Asymptotic Independence of Ω(n)\Omega(n)Ω(n) and Ω(n+1)\Omega(n+1)Ω(n+1) along Logarithmic Averages

More like this

Technical Summary: Asymptotic Independence of $\Omega(n)$ and $\Omega(n+1)$ along Logarithmic Averages