A registered result in its source context
Original ledger classification: Numerical [N] (out-of-sample cross-validation)
Recorded statement: LEAVE-ONE-OUT SEED CROSS-VALIDATION (2026-06-21): turns the 'many observables, one seed phi0' liability into a falsifiable out-of-sample test [N] (Python-only, external data). [N] back-solving phi0 independently from the 5 seed-grammar…
Statement, assumptions and sources
Paper context: Red-team audit
The exact identifier, original classification and full claim are available below.
The source wording and classification below are preserved as recorded. A paper link, shared script or authored topic relation does not upgrade the claim or report a new proof run. Read later reviewed findings separately.
Read the exact registered claim and its dependencies
SEED.CROSSVAL.01
Original ledger classification: Numerical [N] (out-of-sample cross-validation)
Original claim text
LEAVE-ONE-OUT SEED CROSS-VALIDATION (2026-06-21): turns the 'many observables, one seed phi0' liability into a falsifiable out-of-sample test [N] (Python-only, external data). [N] back-solving phi0 independently from the 5 seed-grammar observables (sin^2 theta12, sin^2 theta13, beta, Omega_b, lambda_C; NuFIT 6.0 / ACT DR6 / Planck / PDG centrals) and error-weighted-averaging reproduces the AXIOM seed phi0=1/(6pi)+48 c3^4=0.0531720 to 0.04% (one real seed, not a per-observable fit; chi^2~4.5/5). [N] the precise subset {lambda_C, Omega_b} clusters within 1.5% relative spread. [N] LEAVE-ONE-OUT: every held-out observable is predicted within 3 sigma out of sample; the most-tensioned hold-out is sin^2 theta13 (~2 sigma, the honest pressure point, DATA.SCORECARD.01/v62). [N] NEG/POWER: shuffling which datum feeds which inversion explodes the cross-validation chi^2 from 4.5 to ~1e5 (>100x). A consistency / out-of-sample statistic over repo-documented centrals; phi0 is FIXED by the axioms (the 'fit' is adversarial framing), so this does NOT upgrade any prediction.
Original classification detail
[N] the per-observable back-solve + error-weighted mean = axiom seed (0.04%) + the precise-cluster spread + the leave-one-out out-of-sample residuals + the shuffle power control. A cross-validation statistic; does NOT upgrade any prediction's status.
Dependencies as recorded
ANCHOR.GEN.01; REG.FREEZE.01; DATA.SCORECARD.01
Open the original ledger source, row 402Location recorded in the ledger
These document locations are recorded directly on the ledger row. They are document level navigation, rather than a newly inferred section citation.
- Red-team audit
Sections that explicitly cite this claim
A literal claim identifier is present in these paper sections. The section and its premises determine what the citation supports.
- 2026-06-21 (adversarial-review follow-ups: a structural anti-numerology firewall, an out-of-sample seed cross-validation, an operational data watchdog, and the Target-A holomorphy kill test: v305– v307)Research changelog; TFPT — Introduction and reading guide
Navigation through a shared verification script
These sections cite a script associated with the claim. This is a source navigation join, rather than an explicit citation of the claim or a theorem equivalence.
- Falsifiersinput 4 origin theory
- 2026-07-01 ( v465 — the cross-sector one-parameter seed test: a CMB birefringence angle and a quark Cabibbo angle point to the same seed within [formula in source] , the [formula in source] -independent externally-runnable slice of v306)Research changelog; TFPT — Introduction and reading guide
- Neutrinostfpt 2 standard model
- NeutrinosStandard Model and flavor structure
- OutcomeRed-team audit
- FalsifiersOrigin Theory
- Falsifiersinput 4 origin theory
- Falsifiersorigin theory
All 11 linked sections
- Front matter / support
- 2026-06-22 (external-review pass: a registry bug fix, the NA62 2026 rare-kaon update, CP-phase prose hygiene, and three firewall/visualisation enhancements: v202, v203, v305, v307, v321)
- 2026-07-01 ( v466 — a sixth seed channel from a new measurement sector: the charged-lepton mass ratio [formula in source] back-solves the same axiom seed to [formula in source] , extending the empirical over-determination to five sectors)
Registered verification modules
- v306_seed_crossval.pyRegistered verification module · registry
Read the description recorded in the module registry
Leave-one-out cross-validation of the shared seed phi0: back-solving phi0 independently from the 5 seed-grammar observables (sin^2 theta12, sin^2 theta13, beta, Omega_b, lambda_C; NuFIT 6.0 / ACT DR6 / Planck / PDG centrals) and error-weighted-averaging reproduces the axiom seed phi0=1/(6pi)+48 c3^4=0.0531720 to 0.04% (one real seed, not a per-observable fit); leave-one-out predicts every held-out observable within 3 sigma out of sample (most-tensioned = sin^2 theta13, ~2 sigma); a data<->formula shuffle explodes the chi^2 by >100x (power control)
A registered module is a source reference. Its presence does not assert that the module was executed in this snapshot.