The Coulomb pseudopotential and the optimality of the Einstein spectrum for the linear Eliashberg ceiling
Year: 2026
Type: preprint
Abstract: This preprint asks whether the single Einstein line still maximizes the linear pairing onset of isotropic Migdal-Eliashberg theory when a pairing-only Coulomb repulsion acts through a cutoff that is common to every spectrum compared at fixed coupling lambda and spectral moment M. It records three exact results. First, the ceiling value is Coulomb robust: the repulsion enters the folded stability form as a positive semidefinite rank-one term, so no spectrum at any mu* >= 0 exceeds C(lambda) sqrt(M) (PROVED). Second, the optimizer is not robust: three two-line spectra sharing (lambda, M, mu*, omega_c) with the Einstein line have a strictly larger linear onset at mu*/lambda = 2/3, 3/4 and 1, with temperature gains of at least 2.59, 29.09 and 82.57 percent (CERTIFIED in exact rational arithmetic by two independently written verifiers). Third, below mu*/lambda = 1/2 four further counterexamples, confined to mu*/lambda = 0.4786 to 0.4890 at couplings lambda = 25.6 to 28, show that one half is not a threshold of the model (CERTIFIED), while optimality is proved in three analytic regimes and a two-line reduction makes every comparison two-dimensional (PROVED). Seven certified Einstein brackets are recorded, including the zero-repulsion calibration C(1) in [0.11461945, 0.11461957]. The floating-point phase data that locate the failure are NUMERICAL and are not part of any proof. Whether the Einstein line is optimal for all lambda at mu*/lambda <= 1/2 is left OPEN. Version 2 extends Section 8 with a common-cutoff realizability analysis and corrects the numerical phase table; it withdraws nothing, and no PROVED theorem and no CERTIFIED counterexample or bracket of Version 1 changed. ADDED, PROVED: the exact realizable mask and the necessary and sufficient conditions under which a fixed-mask crossing survives a common cutoff R, with right-continuity in the reduced coupling q and left-continuity in R2 up to the closing endpoint (a new lemma in Section 8.2); the exact two-line difference kernel that isolates the two-line amplitude (a new lemma in Section 8.3); the small-repulsion threshold r0 = 1/64 uniform on lambda in [3/10, 3] for every R >= 0, together with the exact jump-cliff onset tie (a new theorem in Section 8.3); and the second-variation operator with its local strict-maximality criterion (a new theorem in Section 8.4). ADDED, CERTIFIED: the fixed-k=1 crossing at mu*/lambda = 1/2 is unrealizable under any common cutoff for every lambda in {10, 20, 25, 35, 50, 65, 152} over 171 rational boxes, and the lambda = 152 witness (theta = 33333/583333, q_test = 8671/2000) is genuine at fixed k = 1 over 66 boxes yet unrealizable under any common R (a new corollary and table in Section 8.2); the positivity of the second-variation operator on 33 slabs and 20 fixed-k benchmarks (a new corollary in Section 8.4). NUMERICAL: the phase table now reports the correct lowest realizable non-empty mask count at lambda = 10 (k = 2) and lambda = 20 (k = 5) and the crossing 9 q_E^(1) = q_ph at lambda approximately 9.85 instead of 5.85. Version 2 is deposited as a new version of 10.5281/zenodo.24409773, so both versions stay citable. The bounds are on the model quantity T_lin, the linear instability onset of the linearised equations. They are not bounds on a measured transition temperature, they confirm no superconductor, and they make no materials claim. The Coulomb repulsion is a pairing-only coupling: it does not enter the quasiparticle renormalisation. This preprint has not been peer reviewed. AI agents generated the proofs, code and drafts under D. Rens' direction, and separate AI agents carried out the independent reviews.
Author: Rens, Davis
Language: English
Subfield: Condensed Matter Physics
Field: Physics and Astronomy
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W7237353418
