Scoped exclusion
Bounded coefficients do not imply infinitesimal relative bound
In the question: Automated construction of operator maps and sources
Only the general proof step is excluded. The a=1 control is not an admissible elliptic Dirac operator or a P1/P2 countermodel.
Read the exact result, assumptions and evidence
Result
The inference in original source01:397-405 is invalid: P=aA has relative bound |a|. The independent quarter/reflection control A=sigma_x p_theta+sigma_y p_r, P=a I p_r retains the quarter domain; at a=1, A+P on H1 is not closed. Repair: an actual relative bound<1, bounded symmetric perturbation, or a separate theorem for the total elliptic operator and boundary domain.
Required assumptions
- Audit of the general perturbation argument
Given: a selfadjoint first-order anchor with H1 graph norm and a symmetric first-order perturbation with bounded coefficients on the same domain. The quarter/reflection control uses the explicit compact torus operator in section11.2, not a claimed original P1/P2 source.
Original evidence
- PROOF.txt lines 482–517
Source checked at this snapshot
Original location and fingerprint
experiments/theory-contracts/seam-collar-multifield-20261016/PROOF.txt
3537a2b080df0bb8b5d30c73da7dd83dc8e1ae5a6cfea4bb48a8b99f0080dfce
- 002-section-intrinsic-closed-branch-and-canonical-rigid-obje.tex lines 20–28
Source checked at this snapshot
Original location and fingerprint
documents/archive--tfpt-45--source_extracts--01_boundary_kernel_source/sections/002-section-intrinsic-closed-branch-and-canonical-rigid-obje.tex
1de90a4a7379bef804cc7815d72c0f86bb398a5771e9b00997c4bb15e8815f9e
finding:order-one-infinitesimal-bound-inference-fails