The Twist: a no-scale supergravity field for inflation and a thawing dark-energy tail, with pre-registered tests (v5, consolidated)
Year: 2026
Type: preprint
Abstract: Update (14 November 2026): the current version of this work is The Twist v5.3 (DOI 10.5281/zenodo.24226521), with errata and the currently adopted model (TWIST-I). The concept DOI 10.5281/zenodo.24202420 always resolves to the latest version. Please refer to it. Version 5.2 (14 November 2026): v5 plus updated errata. The v5 paper, exploratory note and code are unchanged; read them with Twist_v5.2_errata_2026-11-14.pdf (supersedes the v5.1 errata). Main points: (1) the fifth-force shield fails at one loop for the thawing tail (d(1/α_a)/d ln x = −b_a/4π; seven escape routes tested); (2) adopted reading TWIST-I: the Twist field rolls as in v5 through matter genesis and is then captured at x_E ≈ 27–29.5 by a trapping sector plus an exact weight-3 stabiliser, W ⊃ (T − T_E)(λ M_P χχ̄ + k Y), at negligible entropy cost; Λ still comes from the cliff (n ≈ 124.1); dark energy is w = −1; (3) CMB predictions are Starobinsky-class at N* = 55.7–57.6; (4) the angle is heavy after capture (no axion overproduction); (5) all locked tests since v5 pass, including isocurvature at both coupling readings (margins 1.4× and 4.2×); (6) a free-neutrino-mass fit and one consistency run were still running at publication; results go to the repository. AI-assisted. Version 5.1 (14 November 2026): v5 plus errata. The v5 paper and exploratory note are unchanged; read them with the new file Twist_v5.1_errata_2026-11-14.pdf. Main changes: (1) the fifth-force shield fails at one loop for the thawing tail: a field with K ⊃ −3 ln(X + X̄) moves the Standard Model couplings by d(1/α_a)/d ln x = −b_a/4π, and seven escape routes were tested (two with locked inflation tests); (2) the adopted reading is now TWIST-I, a split model in which the Twist is the inflaton, its field is captured before nucleosynthesis, and dark energy is a cosmological constant (locked rule PREREG_TWISTI); (3) TWIST-I's CMB predictions are Starobinsky-class at N* = 55.7–57.6; (4) new locked tests pass (matter yield 1.25× observed; isocurvature sensitivity 3.70 and 3.68 ≤ 5); (5) updated costs and test calendar (full-fit H₀ 66.8 for the tail; CMB-S4 cancelled; r test via Simons Observatory and LiteBIRD). Earlier records by this author now point here. AI-assisted. Version 5 (13 Nov 2026) replaces version 4 after an independent referee review. It is one consolidated paper, with every number taken from a single parameter table. The changes: The abstract now states the results as they stand. The n_s prediction failed its locked test (NS-SPA: INCONSISTENT). The late-time preference for PURE-TWIST is 1.8σ locally and 0.6–1.0σ after a trials correction. On the full CMB + DESI + supernova data it is worse than ΛCDM (Δχ² = +1.45). The choice μ = 1 and the cliff form are labelled as post-data choices. The fifth-force shield is stated as an explicit assumption. One construction is kept: TWIST-K. The gravitino argument is corrected. The tail-slope and reheating-temperature numbers are corrected. The axial coupling is written explicitly in canonical variables. Growth, w_a and BAO numbers are each given at a stated conditioning. Missing references are added. Material that is not yet settled is in a separate exploratory note: matter genesis, dark matter, reheating, the string-theory search, and the post-data n_s variants. The attached zip archives the code, the pre-registered rule files with their SHA-256 hashes, and the results. AI-assisted (Claude, Anthropic) under the author's direction.
Author: Eldridge, Danny Lee
Language: English
Subfield: Astronomy and Astrophysics
Field: Physics and Astronomy
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W7226886034
