Answers are grounded in this paper and cite their sources.
Lean 4 kernel checks (decide +kernel, no native_decide, Lean core only) of three colourings of the N smallest primes published by D. Bentley (github.com/DuncanBentley/green-tao-bounds, tag v0.2.0, commit a090c3b7; Zenodo 10.5281/zenodo.23854101): 4 colours with pattern 3-3-3-4 for N = 1063, 4 colours with 3-3-4-4 for N = 3013, and 3 colours with 4-4-5 for N = 16953. The colourings and the bounds they give are Bentley's; this upload checks them and states no bound of its own. Each theorem states that the list is exactly the N smallest primes (primality by trial division, proved sound in Lean; every other number in [2, p_N] has a factor witness), that the colour bit masks are pairwise disjoint and their union is the list's mask, and that colour i has no l_i-term arithmetic progression of prime values. The colouring then witnesses grt_k(l_1, ..., l_k) > N in the sense of O. Kullmann (arXiv:1004.0653, Def. 13). There is no Lean definition of grt. Axioms: [propext, Classical.choice, Quot.sound]. The colouring files are MIT-licensed (Copyright (c) 2026 Duncan Bentley) and are included unchanged with their licence text.
No takes yet. Share an insight, caveat, or question.
Advatiya Thalluri (2026) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: