Answers are grounded in this paper and cite their sources.
Mills proved that there is A > 1 such that ⌊A^(3^k)⌋ is prime for every k. The admissible base was later lowered to 2 (Matomäki), 1.946... (Islam) and 20/11 (Järviniemi). Each of these constructions builds a single chain of primes, which forces the exceptional set for primes in short intervals to be small at the scale of a single interval. We observe that a compactness argument (König's lemma) applied to the forest of all admissible chains requires only a power saving in the density of the exceptional set. As a consequence, if the exceptional-set exponent satisfies μ(θ') < 1 for some θ' < 1 − 1/c, then there is A > 1 such that ⌊A^(c^(k−1))⌋ is prime for all k ≥ 1. Using a result of Bazzanella and Perelli this gives every c > 6/5 unconditionally, and the bounds of Gafni and Tao, which rely on the Guth–Maynard zero-density estimate, give every c > 15/13, including the golden ratio. On the Riemann Hypothesis we obtain every c > 1, while Matomäki's single-chain method gives c ≥ φ. Moreover, for every η > 0 there is A such that ⌊A^(k^(2+η))⌋ is prime for every k ≥ 1. Combined with an almost-all result of Matomäki for numbers with at most two prime factors, the same argument shows unconditionally that for every c > 1 there is A with Ω(⌊A^(c^(k−1))⌋) ≤ 2 for all k ≥ 1. For constant ratios c, the whole deduction is formally verified in Lean 4 with Mathlib; code and certificates are available at https://doi.org/10.5281/zenodo.24280530. Submitted to Research in Number Theory. AI use is declared at the end of the manuscript.
No takes yet. Share an insight, caveat, or question.
Luis Fernando Marcelino (2026) studied this question.
Synapse has enriched 8 closely related papers on similar clinical questions. Consider them for comparative context: