# Repository versus published papers: axiom reconciliation Where the repository and the published LF-series papers diverge, the repository is current. The papers were written against earlier states of the formalisation and record several results as imported named axioms. This page states, for each, what is true at HEAD. Verified 2026-08-09 by direct inspection of HEAD and of the pinned `#print axioms` outputs in `CsdLean4/Tests/AxiomAudit/`. ## Findings | Recorded in the papers as an imported axiom | State at HEAD | Where it lives now | Pinned axiom output | |---|---|---|---| | `invariant_measure_uniqueness_CPN` | Absent as an axiom. The abstract axiom was removed on 2026-06-04 together with the abstract `measure_bridge` lemma it served, which nothing downstream used. Its concrete content is a proved theorem, with the name differing in case from the papers. | `Matrix.UnitaryGroup.invariant_measure_uniqueness_cpn`, `CsdLean4/Mathlib/LinearAlgebra/Projectivization/FubiniStudyUnique.lean` | `[propext, Classical.choice, Quot.sound]` (`Tests/AxiomAudit/MathlibStaging.lean`) | | Busch effect-Gleason | Discharged. Proved on 2026-07-21 by Busch's finite-dimensional argument, and the `axiom` declaration removed. | `CSD.LF2.OperationalPackage.effect_gleason_representation`, `CsdLean4/LF2/EffectGleason.lean` | `[propext, Classical.choice, Quot.sound]` (`Tests/AxiomAudit/Foundations.lean`) | | Rank-one density uniqueness | Discharged on 2026-05-18 via `Matrix.PosSemidef.dotProduct_mulVec_zero_iff`. Now a proved theorem. | `CSD.LF2.rankOneDensity_unique_of_certainty`, `CsdLean4/LF2/BornWrapper.lean` | `[propext, Classical.choice, Quot.sound]` (`Tests/AxiomAudit/Foundations.lean`, pinned 2026-08-09) | | `fs_moment_pushforward_uniform` | Discharged on 2026-05-31. Now a theorem of the same name, proved downstream to break an import cycle. | `CSD.LF4.fs_moment_pushforward_uniform`, `CsdLean4/LF4/MomentUniform.lean` | `[propext, Classical.choice, Quot.sound]` (`Tests/AxiomAudit/LF4.lean`) | ## Consequences No `axiom` declaration remains anywhere in `CsdLean4/`. Every result the papers present as conditional on one of the four items above is unconditional at HEAD, in the sense that it depends only on Lean's foundational triple. This is a statement about logical axioms alone. CSD's physical posits are unchanged and are listed in `AXIOMS.md` section 3; they enter as hypotheses on the types and are therefore invisible to `#print axioms` by construction, not by omission. ## Recorded gap, closed `rankOneDensity_unique_of_certainty` carried no `AxiomAudit` pin when this reconciliation was run, so CI could not have caught drift in it. The pin was added on 2026-08-09 in `Tests/AxiomAudit/Foundations.lean`. All four items above are now under CI coverage.