Эксперимент: ISS-Constrained Modular Recurrent Network (#371)

{ "worked": true, "confidence": 9, "verdict": "Built and ran an ISS-constrained modular recurrent MVP with scalar Euler verification and a small sequence-sum comparison against a vanilla GRU. The mechanism manifested: the observed stability transition occurred at the predicted gamma=2.0 boundary, decay slopes matched log|q| with relative error below 6e-16, perceptual decay matched log(alpha) to numerical precision, and forced steady-state values agreed with the ISS bound within 0.66%. The modular model did not beat the GRU on this task, so the result supports the stability mechanism but not an accuracy win.", "metrics": { "baseline": "Vanilla GRU: 1,969 parameters, final train MSE 0.0502, test MSE 0.03886.", "idea": "ISS modular residual model: 1,129 parameters, final train MSE 0.0890, test MSE 0.05825. Toy verification: gamma=2.0 gives the |q|=1 boundary; gamma=2.1 contracts with |q|=0.99; observed decay slopes match predictions; ISS steady-state relative errors are 0.66%, approximately 1e-15, and approximately 4e-16 for gamma 2.1, 3, and 5." }, "how_to_run": "/home/maxwelhelp/main/bin/python3 iss_experiment.py", "files": [ "iss_experiment.py", "results.json" ], "limitations": "The mechanism verification is an analytically matched scalar linear Euler system rather than a full Jacobian sweep of a trained high-dimensional network. The learning comparison used one fixed seed, a short synthetic sequence-sum task, CPU execution after shared-CUDA allocation errors, and did not test Sequential MNIST, matched FLOPs, multiple seeds, or long-horizon generalization." }