DMDE Triplepoint + CosmoGate: Five-Channel Weak-Rate Closure and Frozen-Pair Differential-Response Design (v0.9.20)

Status: working hypothesis; methods/software checkpoint; not peer reviewed; no detection or production BBN-result claim.

DMDE v0.9.20 is a source-frozen validation and numerical-design release for a blinded two-card neutrino-spectrum to Big Bang nucleosynthesis (BBN) test. It repairs a reproduced synthetic v0.9.19 false pass in which both charged-lepton capture channels were omitted and unblocked vacuum beta decay was returned at all temperatures. The new contract independently recomputes audit-only, neutron-lifetime-normalized Born histories for all five exported weak processes, applies per-process and direction-total envelopes, and checks the relevant threshold and endpoint refinement. It retains source-shape and time-integral closure, source-on/off production-stepper witnesses, and BBN input-consumption canaries. A companion source-only result derives the frozen pair's exact differential-response geometry and conditional numerical-precision requirements. The two physical source payloads and model parameters are unchanged relative to the byte-frozen v0.9.19 parent. The public packet contains the blinded provider files, validators, schemas, templates, documentation, and receipts. It contains no production helium, deuterium, or effective-neutrino-number result.

What changed

• All five exported weak-process histories receive independent Born-envelope checks: neutrino capture, antineutrino capture, positron capture, electron capture, and neutron beta decay.
• Each non-negligible process has a 25% audit envelope; each neutron-to-proton or proton-to-neutron direction total has a tighter 10% envelope. Finite-temperature beta blocking and the late neutron-lifetime gate remain explicit.
• The momentum grid must meaningfully refine the antineutrino-capture threshold at 1.80433131 MeV and the beta-decay endpoint at 0.78233341 MeV.
• Earlier source-shape, source-time, transport-stepper, meaningful-refinement, and BBN-consumption evidence requirements remain in force.
• The source-only frozen-pair calculation gives an exact response identity and makes the numerical precision needed for future endpoint comparisons explicit.

Scientific status and limits

• No production Y_p, D/H, or N_eff prediction is bundled or implied. No confirmation of DMDE or a unified theory follows from these checks.
• The Born envelopes are execution checks, not a complete corrected weak-rate calculation. The frozen five-column export does not separately represent the inverse three-body process p + e- + antineutrino -> n.
• Passing cannot prove that evidence was not fabricated or that transport, radiative, recoil, plasma, or nuclear-network physics is complete. Independent replay, physical review, and backend replication remain necessary.
• The pair calculation determines a direction in source-amplitude space. It does not determine the sign or magnitude of any abundance response. Its neutrino coordinate includes linked source-energy and parent-density effects.

Files and verification

The original v0.9.20 provider archive is preserved byte for byte. The public README and release-note PDF explain the scope and next evidence required. The public verification receipt distinguishes software and archive checks from missing production results. The SHA-256 manifest covers the public files. Private analyst scoring and unblinding material is excluded.

Source release: 17 August 2026. Explanatory note prepared: 5 September 2026. This is a methods/software checkpoint with production execution and independent replication still pending.