Prime values of Ramanujan's tau function
Boyuan Xiong
TL;DR
The paper investigates how often primes occur as values of Ramanujan's $\tau$-function. It combines Hecke multiplicativity, modular congruences modulo $23$, and Diophantine tools (Schinzel-type bounds and Bombieri–van der Poorten estimates) to control $\tau(p^k)$ for prime powers. It proves that for many residue classes mod $23$, the primes that are $\tau$-values are extremely sparse, and in particular there exists a set of primes of Dirichlet density at least $\frac{9}{11}$ that are not $\tau$-values, while the set of primes that are $\tau$-values has density zero. The results complement questions around Lehmer's conjecture by focusing on the image of $\tau$ rather than its zeroes, and showcase a sharp dichotomy between prime values and omitted values in the $\tau$-orbit.
Abstract
We study the prime values of Ramanujan's tau function $τ(n)$. Lehmer found that $n=251^2=63001$ is the smallest $n$ such that $τ(n)$ is prime: $$τ(251^2)=-80561663527802406257321747.$$ We prove that in most arithmetic progressions (mod 23), the prime values $τ$ belonging to the progression form a thin set. As a consequence, there exists a set of primes of Dirichlet density $\frac{9}{11}$ which are not values of $τ$.
