The Chowla conjecture is a famous open problem in number theory about the randomness of the Liouville function, which assigns +1 or -1 to integers based on whether they have an even or odd number of prime factors. Specifically, the conjecture predicts that the values of this function at different integers should be essentially uncorrelated with each other, behaving like independent coin flips. A logarithmically averaged version of this conjecture, where you weight integers by 1/n rather than treating them equally, was proved by Tao in 2016, but without concrete estimates on how large the numbers need to be before the cancellation becomes visible.
This paper makes Tao's proof quantitative. The authors work out explicit bounds on how large the averaging range needs to be before you can guarantee that the Liouville function at two shifted inputs shows the predicted cancellation. The answer involves tower-exponential functions, meaning the required range grows extremely rapidly with the parameters involved, but the key contribution is that these bounds are now uniform across different choices of the shifts and windows being averaged over. The paper also handles two interleaved inputs of the form involving linear functions of n with integer coefficients, and tracks carefully how the required threshold depends on those coefficients. A secondary result gives fully explicit numerical bounds relating to how "pretentious" the Liouville function can be, meaning how closely it can mimic a structured multiplicative character, with conditional improvements if the Generalized Riemann Hypothesis is assumed.
The technical machinery draws on several tools from analytic number theory: explicit zero-free regions for Dirichlet L-functions, Harnack's inequality from complex analysis, and careful comparisons of Euler products (infinite products over primes that encode multiplicative structure). While the bounds produced are far too large to be computationally useful in practice, the work is significant because it transforms an existence proof into a concrete statement, opening the door to future improvements and providing a cleaner foundation for understanding when and why multiplicative functions behave randomly.