A problem of Polya
Umberto Zannier
TL;DR
The paper analyzes a non-constant-coefficient linear recurrence tied to a carefully chosen rational sequence, uncovering unexpected $p$-adic congruences and denominators that are powers of 2. It reveals that the generating function is algebraic and intimately connected to an elliptic curve, enabling a p-adic and differential-analytic treatment via the Cartier operator and logarithmic derivatives. By leveraging Grothendieck’s conjecture, Honda–Katz theory, and the Chudnovski–André theorem, the authors establish congruence properties (Theorem T.congr) and a converse result (Theorem T.converse), and they relate these to integrality and p-curvature phenomena. The work ties Recurrence-Arithmetic to algebraic geometry, using explicit elliptic-curve models to illustrate and prove the key congruences, while discussing Polya-type questions and density issues for primes, with several open questions and CM-specific observations.
Abstract
We prove an improved form of an expectation of Polya and discuss several related questions
