Answers are grounded in this paper and cite their sources.
A scalar finite-volume criterion for a continuum mass gap in SU(2) lattice gauge theory (note v1.0, 12 November 2026; first deposit). What the note is. A technical note that transposes the two-dimensional finite-volume doubling criterion of OpenAI (Proposition 2.2 of “Sharp mass bounds for the two-dimensional O(4) model”, 23 September 2026) to lattice gauge theories in dimensions four and three. It is predominantly a transposition, not a new mass-gap mechanism; no priority is claimed for the mechanism. Results. For the SU(2) Wilson action on periodic tori, set Δβ(n) = 16 log Z(n4) − log Z((2n)4). Under hypercubic symmetry and a positive trace-class transfer representation (a hypothesis, not formalised), Δ(n) ≤ 2−9 implies Δ(2n) ≤ 256 Δ(n)2 and (λi/λ0)n ≤ 256 Δ(n) ≤ 1/2 on every spatial torus (2jn)3 (in dimension three: 2−7 and 64). If Δβk(nk) ≤ η ≤ 2−9 along lattices of spacing ak = ℓ/nk and the lattice Osterwalder–Schrader forms converge (the existence of the continuum limit, assumed), the limit satisfies the q-form of the gap with q = 256η; under the Osterwalder–Schrader reconstruction (also assumed) its mass gap is at least −log(256η)/ℓ ≥ (log 2)/ℓ and its vacuum is unique. This implication is formalised in Lean 4 (35 declarations, standard axioms only) against the author’s own Lean statement of the problem (ClayYM v1.1, doi:10.5281/zenodo.23644079), with the transfer representation, the convergence and the reconstruction as named hypotheses. A converse (finite-volume gap including electric-flux sectors, plus a trace bound) is proved on paper at a fixed physical size, with counterexamples when either ingredient is dropped. Status. The Lean sources were rebuilt on 11 November 2026 by a non-author on a clean machine from hashed copies, and again from the git archive of this package (see the note, Appendix A). The torus–cylinder comparison, the converse and the witnesses are proved on paper only. The strong-coupling instance (Δβ(2) ≤ 1.741·10−3 for β ≤ 0.005) is conditional on a finite enumeration and a rational comparison reported from one program and not independently certified. The Maxwell witness is conditional on an identification with a normalised Villain measure that is not derived; the massive-scalar thresholds are numerical. Written by AI models and read by other AI models; no human has checked the proofs; not refereed. Relation to earlier public work. Version 5.0.0 of the series of Bhattacharjee et al. (The Yang-Mills Mass Gap from One Bound at One Scale, doi:10.5281/zenodo.24147967; record created on 11 November 2026, before this deposit, and read by us after this note was written) contains, for four-dimensional tori, the same doubling defect, its bracket by thermal traces, the purity reading, and a one-scale criterion whose hypothesis, a family of inequalities between partition functions of finite tori, implies ours and whose conclusion is wider. For the Wilson theory in dimension four, the lattice part of this note is therefore a variant of their criterion with a weaker hypothesis and a narrower conclusion (dyadic cubes only). No priority is claimed for these items (Section 1.4 of the note, which also lists what we did not find in the passages read). Limitations. No estimate of Δ at weak coupling, no existence or non-triviality of the continuum limit, no reconstruction. The criterion does not apply as it stands to the formalised construction of OpenAI (Section 9 of the note). Reading Δ as a combination of Rényi-2 entropies of the thermal weights is a reformulation, not a result. The criterion is a sufficient condition; its equivalence with the Yang–Mills mass-gap problem is not claimed. The anteriority search is limited and recorded in the note. Prepared with the assistance of Claude (Anthropic: Fable 5.2, Opus 5.5, Sonnet 5.5), GPT-6.1 Sol and GPT-6 Astra (OpenAI), Qwen3.8 (Qwen team) and DeepSeek V4 Pro (DeepSeek); the roles are listed in the note.
No takes yet. Share an insight, caveat, or question.
Kevin Remondière (2026) studied this question.
Synapse has enriched 8 closely related papers on similar clinical questions. Consider them for comparative context: