Answers are grounded in this paper and cite their sources.
We prove an upper bound on the linear onset temperature of a reciprocal multichannel Eliashberg model whose channels carry different bounded, energy-dependent, even densities of states and a common retarded glue. The bound reads kBTlin < C sqrt(rho), where rho is the norm of the glue moment matrix, and the constant is C = C* = sqrt(341/250)/(2 pi), which rounds to 0.1858777115, for the supremum over the class, and C = Crep = sqrt(6821/5000)/(2 pi), which rounds to 0.1858913384, as a strict bound for every model of the class. Both numbers lie between the exact scalar constant Cinf = 0.18272624777232952 and the constants sqrt(683/500)/(2 pi) = 0.1860139355 and sqrt(41/30)/(2 pi) = 0.1860593214 of the preceding preprint, and both are certified by exact rational arithmetic. The relative excess over Cinf falls from 1.82% to 1.73%, and the new constant sits 0.25% above the floor 0.185411 that the same edge-reserve method cannot go below. The gain comes from a reserve raised from 317/1000 to 159/500, a finer partition of the head weights, and a minimisation over the exact rational tangent envelope in the inverse head weight. The finite part of the argument is a covering of a compact rectangle in two variables by closed rational rectangles, carried out twice by independent programs with 159 and 466 leaves and guarded by twelve deliberately corrupted runs. The result is an upper bound on a linear model onset temperature. It is not an exact transition temperature, it is not a measurement, and it makes no statement about any material. The question whether the supremum over the class equals Cinf remains open. The floor 0.185411 is a limit of the edge-reserve method for the certified weight sequences, not a lower bound on the supremum. The replay supplement checks the finite exact-rational coverings and their guarded mutants only; it does not formalize the analytic arguments. This preprint has not been peer reviewed. AI agents generated the proofs, code, and manuscript drafts under D. Rens' direction. Separate AI agents performed the independent reviews.
No takes yet. Share an insight, caveat, or question.
Davis Rens (2026) studied this question.
Synapse has enriched 8 closely related papers on similar clinical questions. Consider them for comparative context: