Oscillation of partial sums of the Möbius function and zeros of Riemann's zeta function

In the vast landscape of mathematics, there is a single question that has haunted scholars for nearly two centuries: how are prime numbers distributed? Primes are the building blocks of arithmetic, the indivisible atoms from which all other whole numbers are constructed. While they seem to appear at random, mathematicians have long suspected a hidden order beneath the chaos. This order is encoded in a complex mathematical object known as the Riemann zeta function. The behavior of this function, specifically the location of its "zeros"—points where the function's value drops to zero—dictates the rhythm of the primes. If these zeros lie on a specific line, the primes follow a predictable pattern; if they stray, the pattern becomes erratic. For over a hundred years, the connection between these zeros and the errors in our predictions of prime numbers has been a central mystery, driving some of the most profound work in number theory.

A recent paper by mathematician János Pintz tackles a specific, difficult aspect of this mystery: the behavior of the partial sums of the Möbius function. To understand this, imagine a running tally where you add or subtract one for every number based on its prime factors. This tally, known as the Mertens function, is a sensitive barometer for the distribution of primes. For decades, mathematicians have tried to predict how high or low this tally can swing. A famous conjecture once suggested the tally would never exceed the square root of the number being counted, but this was proven false in the 1980s. The question that remained was not just about the size of these swings, but about their precise relationship to the zeros of the zeta function. Does the average size of these swings match the size of the largest single swing? And do both of these match the theoretical maximum predicted by the zeros?

Pintz's work provides a definitive answer to these questions, establishing a rigorous link between the erratic behavior of the Möbius function and the geometry of the zeta function's zeros. The paper proves that the average magnitude of the Möbius function's swings and the maximum magnitude of those swings are, in the long run, asymptotically equivalent in their logarithmic order. More importantly, the author demonstrates that both of these values are determined with high accuracy by a real function derived from the distribution of the zeta function's zeros. In simpler terms, the paper shows that the "noise" of the prime numbers is not random chaos but is tightly controlled by the position of the most dominant zero. This finding unifies two different ways of measuring the error in prime number predictions—the average error and the worst-case error—showing they grow at the same rate and are governed by the same underlying force.

The research builds upon a century of effort, refining earlier results that had only established loose connections or worked under strict assumptions. Previous work had shown that if the zeros of the zeta function stay within a certain region, the error in prime counting stays small. However, the reverse question—whether the size of the error tells us exactly where the zeros are—was much harder to solve. Pintz's paper addresses this by showing that the growth rates of the average and maximum errors are asymptotically equivalent to the growth rates predicted by the zeros. By developing new techniques to handle the mathematical singularities that arise when working with the reciprocal of the zeta function, the author was able to prove that the relationship is a precise asymptotic equivalence in terms of logarithmic order, meaning the ratio of their logarithms approaches 1. The paper confirms that the maximum value the Möbius function reaches in a given range and the average value it maintains over that same range share the same logarithmic growth rate. This means that as we look at larger and larger numbers, the difference between the typical behavior and the extreme behavior vanishes in a logarithmic sense, even if the values themselves do not become identical.

A crucial part of this achievement involves dealing with the case where the zeros might lie very close to the critical line that separates order from chaos. The paper addresses the scenario where the zeros are as far to the right as possible without breaking known mathematical laws. In this difficult case, the author proves that the average and maximum values still converge to the same prediction derived from the zeros. This result is significant because it removes the need for complex, abstract regions of "zero-free space" to make these predictions. Instead, the paper shows that one can simply look at the sum of the contributions from all the zeros to understand the behavior of the function. The author establishes that the sum of these contributions, the maximum single contribution, and the actual observed values of the function all grow at the same rate.

The paper also clarifies the relationship between the Möbius function and the error term in the prime number theorem, which counts the number of primes up to a certain point. While these are two different mathematical objects, Pintz shows that their oscillatory behaviors are nearly identical. The maximum deviation and the average deviation for the prime counting error and the Möbius sum are shown to be governed by the same function of the zeta zeros. This implies a deep structural unity in how the primes and their related functions fluctuate. The work does not rely on unproven hypotheses like the Riemann Hypothesis itself; rather, it holds true regardless of where the zeros actually lie, provided they are within the known bounds. The results are proven mathematically, not simulated or suggested, offering a solid foundation for understanding the limits of prime number distribution.

In the end, this paper resolves a long-standing ambiguity about the nature of these mathematical oscillations. It confirms that the "worst-case" scenario for the distribution of primes is not significantly worse than the "average" scenario when viewed through the lens of logarithmic growth. Both are dictated by the same underlying geometry of the zeta function. The work serves as a powerful confirmation that the seemingly random fluctuations of prime numbers are, in fact, a highly structured phenomenon, tightly bound to the location of the zeros of the Riemann zeta function. By proving that the average and maximum orders of magnitude coincide in their logarithmic growth, the paper provides a clearer, more precise map of the terrain where the primes live, showing that the extremes and the everyday are two sides of the same coin.