HLV-LIRO SIM1 v2.0.0: Locked Rigorous Contract and Invariance Audit of the HICE0–HICE6 System — Protocol, Software, and Results

This record contains the locked protocol, executable software, one-click Google Colab notebook, and complete primary results for Simulation 1 of the HLV Layered Identity-Routed Operator Architecture (HLV-LIRO) programme. SIM1 is a synthetic contract-and-invariance audit rather than a comparative architecture-performance benchmark. Its primary question is whether the declared HLV Identity-Channel Evaluation contracts HICE0–HICE6 are internally coherent, numerically testable, mutation-sensitive, and correctly localized under controlled graph variation, admissible and inadmissible transformations, bounded perturbations, selective memory ablations, cycle-order controls, context–content leakage probes, and matched-null construction attempts. The locked v2.0.0 run used 24 sealed seeds, six graph families, two graph-size tiers, and 1,440 primary base graphs. It was executed deterministically in float64 and included explicit type, interface, backbone, gauge, perturbation, memory, cycle-order, leakage, and null-model mutation controls. Statistical evaluation used 2,000 seed-level bootstrap resamples, 1,000 hierarchical-bootstrap sensitivity resamples, 10,000 sign-flip permutations, 95% confidence intervals, and Holm–Bonferroni multiplicity correction. The executed configuration and scientific source exactly matched the locked SHA-256 commitments. All 1,440 graph tasks completed without unhandled errors, and every file listed in the result manifest passed independent SHA-256 verification. Total runtime was 1,464.38 seconds on a two-CPU Google Colab environment. HICE0 type/interface integrity, HICE1 exact backbone fidelity, HICE2 local U(1) gauge covariance, HICE4 selective long-horizon memory dependence, the noncommutative cycle-order witness, and the preregistered seed-level HICE5 context–content separation gate passed. Two gates failed. HICE3 failed because the maximum inside-radius normalized spectral drift was 1.715 × 10^-3, exceeding the locked 10^-3 tolerance. The violation was localized to random-geometric graphs; adjacency preservation, forced-threshold-crossing detection, and outside-versus-inside sensitivity controls passed. HICE6 failed because the locked degree-preserving null-construction procedure achieved admissible coverage of only 0.497 (95% CI 0.490–0.500), below the required lower-bound threshold of 0.80. Acceptable matched nulls were obtained for random-regular, grid, and nearly all scale-free targets, but not for small-world, random-geometric, or crystal-stress targets. The preregistered overall verdict was therefore: CONTRACT_SYSTEM_INCONSISTENT with local tags: STABILITY_RADIUS_NOT_LOCALIZED MATCHED_NULL_CONSTRUCTION_FAILURE These failures are restricted to the tested SIM1 software contracts. They do not validate or falsify HLV ontology, physical time, biological memory, consciousness, LIRO performance superiority, or HLV-specificity. Performance comparison belongs to SIM2, and holonomy-memory specificity belongs to SIM3.