SFM 2.1 Computational Verification Companion I: Verification Before Validation — Exact Construction of the Admissible Mapping Family T_a and Six-Case Verification
This technical companion provides a finite small-N construction of the admissible mapping family T_a introduced in SFM 2.1 Version 1.1. The construction compares two consecutive labelled relational graph states rather than computing the automorphism group of one state against itself. Candidate datum bijections are enumerated exactly and filtered by carrier closure, datum-origin preservation, locked-topology preservation, reachability, and minimal AIT exclusions. Surviving mappings are scored by a weighted root-mean-square distance over locked relation class and token lineage. Direction remains assigned to ordering, while relation magnitude remains assigned to separation and uniformity. This factorised distance is one declared operational specialisation of the open d_K interface, not the only possible persistence metric. The implementation is verified against the six published worked cases. The baseline returns P = 1; the relation-class perturbation returns P = 0.591752 and fails at mu = 0.75; the ordering, separation, and uniformity perturbations retain P = 1 and remain isolated to their intended audits; and carrier loss returns undefined before persistence is evaluated. The supplied reproducibility package contains the implementation, regression tests, six worked cases, complete candidate-mapping results, integrity hashes, and execution instructions. An independent external AI computational reproduction audit classified Companion I as reproduced without material error. All five regression tests passed, all six worked cases matched exactly, and all 40 candidate-mapping records were reproduced. This is computational verification of the declared finite formalism. It is not empirical or physical validation of a pre-Planck substrate. SFM 2.1 Version 1.1 remains frozen and unchanged.
Authors
- A J Paton
Publication Details
- Journal
- Zenodo (CERN European Organization for Nuclear Research)
- Published
- 2026-07-17
- DOI
- https://doi.org/10.5281/zenodo.21410613
- Primary Topic
- Scientific Computing and Data Management
- Type
- preprint