Figure X — Integrated Empirical Validation of the Collapse Criterion and Geometric Inference Failure
Craig Kyrle Strachan Davidson
This dataset presents the full empirical realisation of the Forensic Inversion Protocol (RE-INVERT), validating the Collapse Criterion governing inference failure in multi-witness systems. The figure integrates six panels (A–F), each corresponding to a structural component of the underlying geometric inference framework.Panel A demonstrates accurate recovery of interaction amplitudes under adjoint reconstruction, confirming numerical stability even near collapse. Panel B quantifies the competition between geometric stability, via the minimum eigenvalue of the Fisher Information Metric λ_min(t), and cumulative environmental distortion D(t). Panel C tracks the evolution of the dimensionless collapse parameter Γ(t), showing smooth convergence toward the critical threshold. Panel D empirically confirms loss of identifiability as Γ → 1, with reconstructed states exhibiting extreme dispersion. Panel E analyses the role of regularisation ε, demonstrating that bias–variance trade-offs cannot compensate for geometric degeneration. Panel F provides regime classification across rigid, stable, transitional, and collapsed phases.The results establish that inference failure arises from geometric degeneration of the Fisher metric, rather than numerical instability or noise amplification. The collapse parameter Γ(t) emerges as a universal control variable governing the transition from identifiable to non-identifiable regimes.This dataset supports and extends the theoretical results presented in:Trace Forensics — Paper 31: Fisher-Geometric Inference under Cascade RefinementTrace Forensics — Paper 2: Quantum Forensics via Environmental Bow WavesTrace Forensics — Paper 5: Multi-Witness Inference Foundations
•Apr 21, 2026•0 citations