/teal-sea
teal-sea / zeta-labstate of record · compiled 28 Sep 2026 · revision e4945c4 · source

Library · lean/ZetaLean/Pub1/OBLIGATIONS.md

Pub 1 strong closure, status: CLOSED

481 words · 78 lines · source

Toolchain leanprover/lean4:v4.33.0-rc2, Mathlib v4.33.0-rc2 (51e6992efd06126df61a496bebf8f49482a4e129).

Both headline theorems are unconditional.

pub1_strong_closure :
  IsProfile w → ∃ C₁ C₂, IsLUB (quot '' sourceAdmissible C₁ C₂) (cStar w)

pub1_strong_closure_reciprocal :
  IsProfile w → ∃ C₁ C₂, IsGLB (recipQuot '' sourceAdmissible C₁ C₂) (cStar w)⁻¹

pub1_strong_closure_exists :
  ∃ w C₁ C₂, IsProfile w ∧ IsLUB … ∧ IsGLB …

The only hypothesis is IsProfile w, which is the setting: it says w solves Aw = 1, the profile the statement is about. exists_profile proves such a w exists and pub1_strong_closure_exists carries no hypothesis at all, so nothing is vacuous.

lake build succeeds; no sorry, no admit, no axiom declaration, no native_decide; all audited theorems depend only on propext, Classical.choice, Quot.sound.

How the four obligations closed

A, F₁'' = 2δ₀ + q and interior C² of w. pub1_kernel_second_deriv : (Tv)''(s) = 2·v s + ∫_I q(|s-t|)·v t dt, with the delta mass obtained as the two moving-endpoint terms of the integral split at t = s, each contributing fKer'(0)·v(s) = v(s). No distribution theory was built. Then w_contDiffOn : ContDiffOn ℝ 2 w (Ioo (-1/2) (1/2)), |w''| ≤ 2 + 80.963…, |w'| ≤ B₂/2. Claimed on the open interval only: w'' genuinely jumps by 2w(±1/2) ≥ 2/5 at the endpoints.

B, the exact residual certificate. r0_identity : u + T₀u + r₀ = 1 on I (M = 20, degree-52 rational residual, 132 atomic integrals), the exact ∫_I r₀², both series tails, ‖q‖_∞ ≤ 80.963, ‖q - q_M‖_∞ ≤ 2.90e-17, the two resolvent estimates, and finally

zpp_bound : |w''(s) - u''(s)| < 6060899845/10^12      (s in the interior)

via the certificate identity z'' = r₀'' - 2z - q*z - (q - q_M)*u, which is the difference of w'' = -(2w + q*w) and u'' = -r₀'' - (2u + q_M*u). The final arithmetic closes with margin 3.6e-13.

C, positivity. profile_pos : ∀ s, 0 < w s.

D, uniform L¹ bounds. taper_secondDeriv_L1_of_profile and taper_sq_secondDeriv_L1_of_profile, uniform for L ≥ 8.

Strict concavity and radial monotonicity. w_second_deriv_lt : w'' < -0.59326318 on the interior, hence w_deriv_neg, w_antitoneOn, and w_radial : |s₁| ≤ |s₂| ≤ 1/2 → w s₂ ≤ w s₁.

w_radial supersedes Concavity.radial_antitone, which assumed Differentiable ℝ w on all of ℝ. w does not satisfy that at ±1/2, so that form was uninstantiable; the replacement needs only interior differentiability plus continuity on the closed interval, via antitoneOn_of_deriv_nonpos. Concavity.lean is retained for its abstract even/concave lemmas.

The lesion

Pub1LesionTest.lean.expected-fail must not compile, and does not: with the residual bound inflated by 101 the arithmetic step reduces to False. Its kernel-checked counterparts are concavity_margin_lesioned and, on the final assembly, zpp_arith_lesioned / concavity_lesioned_final.

Provenance

ZetaLean/Pub1/Aristotle/ came from Harmonic's Aristotle (ids in lean/ARISTOTLE-RUNS.md). Per lean/proof_adapter.py the service's own claim is input, not evidence: each passed the static refusal scan and lake build here before use.