Documentation

CsdLean4.LF4.DuistermaatHeckman

LF4: the Duistermaat–Heckman pushforward (qubit instance) — TOMBSTONE #

Category: 3-Local (the Duistermaat–Heckman pushforward (qubit instance) — TOMBSTONE).

This file is a tombstone. It introduces NO axiom. It formerly named, as a Mathlib-external axiom fs_moment_pushforward_uniform, the qubit Duistermaat–Heckman / Archimedes hat-box fact

(fun p => momentMap p 0)∗ fubiniStudyMeasure p₀ = volume.restrict [0,1]

(the N = 2 instance: the moment map pushes the Fubini–Study measure on ℂℙ¹ to Lebesgue on the polytope [0,1]; equivalently |U₀₀|² ~ Uniform[0,1] for Haar U(2)). That axiom was discharged 2026-05-31 — it is now a theorem of the same name, proved in CsdLean4/LF4/MomentUniform.lean via the Gaussian route (Gaussian-induced μ_FS + moment-marginal change of variables), alongside the foundational-triple-only fs_born_volume_ratio_qubit_uncond and qubit_born_frequency_convergence_uncond. The general-N extension (fs_moment_joint_dirichlet_N, MomentDirichletN.lean) is likewise a theorem.

The content moved downstream to break an import cycle (the discharge needs GaussianCP / MomentRatioUniform, which would cycle if proved here). The honest-boundary framing it once carried — that this is a pure geometry statement about μ_FS and the moment map, in which the Born rule does not appear, so the qubit Born value is genuinely derived from the Kähler structure — now holds unconditionally, since the geometric input is proved rather than posited. See the dated note below and specs/carve-out-plan.md Tranche M (plan B, CLOSED).

Discharged (2026-05-31): no longer an axiom. #

The Duistermaat–Heckman / Archimedes fact for the qubit — (momentMap · 0)∗ fubiniStudyMeasure p₀ = volume.restrict (Icc 0 1) — was carried here as axiom fs_moment_pushforward_uniform. It is now a theorem of the same name, proved in CsdLean4/LF4/MomentUniform.lean (plan B: Gaussian-induced μ_FS