Audit panel A: Bernstein approximation and prior-art collision

Objective

Independently determine whether the final V4 Bernstein-cone contribution is new and correctly distinguished from prior work.

Frozen V4 target

• branch: review/bernstein-obstacle-v4-final
• commit: 61594952ad880d2b61759cfa93a19df979183c09
• analytical integration PR: #114
• parent audit: #96

Earlier v1/v2/v3 targets remain available only for provenance.

Claims to test

1. Nonnegative Bernstein coefficients are a known sufficient global range certificate and are not claimed as new.
2. Bounds-constrained high-order FEM and obstacle methods are known and are not claimed as new.
3. The candidate novelty is the combination of:
   • a conforming Bernstein coefficient obstacle cone;
   • Mosco recovery for that cone;
   • threshold-form constructive scheduling of smooth/FEM recovery stages;
   • conformity-preserving global coefficient clipping near a regular free boundary;
   • the corrected conditional h^r + h_Gamma^(3/2) energy estimate under explicit local grading, regularity, multiplier, and boundary hypotheses;
   • certified adaptive/contact computation;
   • a pinned Lean finite/abstract bridge with an explicit analytical trust boundary.

Required comparison set

• Allen–Kirby bounds-constrained Bernstein approximation;
• Kirby–Shapero high-order bounds-satisfying FEM;
• higher-order primal and mixed obstacle FEM;
• proximal Galerkin and other structure-preserving pointwise-constraint methods;
• hp/spectral obstacle methods with GLL constraints;
• certified polynomial range-bound methods;
• formal Bernstein/de Casteljau work in proof assistants;
• any earlier Bernstein–Bézier obstacle/contact discretization or Mosco analysis.

Deliverable

A source-cited report stating exact collisions, exact claims that remain new, missing references or terminology, whether the novelty language is acceptable, and the search strategy/databases used. A negative result or prior-art collision is fully eligible.

Internal scope label

FUN BUX: 3,500. This is an unfunded internal effort/complexity label, not a real or promised bounty. No payment is represented as available unless separate written funding terms are posted.

Inspect the frozen V4 target at commit 61594952ad880d2b61759cfa93a19df979183c09, along with analytical integration PR #114 and parent audit #96. Compare the required prior-art categories, then produce a source-cited report identifying exact collisions, remaining novelty claims, missing references or terminology, and the search strategy and databases used.