The exact degree of divisor-sum identities for the prime-counting function modulo prime powers
Polymath4 computed the parity of pi(x) below the square-root barrier, but its approach to pi(x) modulo 3 stalled at the cost of the three-fold divisor sum D_3(x). We show that, within a natural family of divisor-sum identities, a three-fold divisor sum is forced, and we determine the exact degree needed for each prime power. For the target that yields pi(x) modulo p^k, the least possible degree bound is exactly S(p^ceil(k/2)), and exactly S(p^k) without inverse factors, where S is the Kempner function; both values are attained by explicit constructions, over the integers, every ring of integers of a number field, and Z_p. We also prove that pi(x) mod 3 and D_3(x) mod 9 determine each other, and give an algorithm for pi(x) mod 3 proved correct for every x. Every theorem in the paper is machine-checked in Lean 4 with Mathlib.
Authors
- Joseph S. Farland (ORCID: https://orcid.org/0009-0007-6809-804X)
Publication Details
- Journal
- Zenodo (CERN European Organization for Nuclear Research)
- Published
- 2026-11-21
- DOI
- https://doi.org/10.5281/zenodo.24408922
- Primary Topic
- Analytic Number Theory Research
- Type
- preprint