Эксперимент: Trajectory-Certified Contractive RNN (#367)

{ "worked": true, "confidence": 8, "verdict": "Built a scalar Trajectory-Certified Contractive RNN MVP with the lifted sector LMI, trajectory-based identification, and model/data certificate comparison. The mechanism manifested: measured local contraction matched the predicted |a+b| essentially exactly, certification transitioned near the predicted boundary (observed 0.489 vs predicted 0.55, 11% error), and disturbance response grew sharply toward and beyond the boundary. This demonstrates stability checkpoint selection, but not improved task accuracy.", "metrics": { "baseline": "Uncertified recurrence at a=0.60: measured local perturbation ratio 1.045 and finite-horizon impulse response l2=10.035; hidden perturbations grow.", "idea": "Certified region included a=0.50 with ratio 0.950 and impulse response l2=2.251, while certification failed at a=0.60; model-based and trajectory-derived certificates agreed on sampled points. Certificate gamma values were conservative, e.g. 61.054 at a=0.50." }, "how_to_run": "/home/maxwelhelp/main/bin/python3 certified_rnn_experiment.py", "files": [ "certified_rnn_experiment.py", "results.json", "REPORT.txt" ], "limitations": "CVXPY was unavailable, so the general SDP was approximated in this scalar MVP by a bounded grid over P, lambda, and gamma^2 with direct LMI eigenvalue checks. No multidimensional RNN, noisy-state reconstruction, training task, accuracy comparison, or large-scale computational benchmark was tested." }