# Эксперимент: Krylov Resonance Regularization (#1143)

{
  "worked": false,
  "confidence": 9,
  "verdict": "Built a readable MVP with matrix-pencil/Krylov pole estimation, Hankel conditioning diagnostics, a controlled correlation sanity check, and a delayed-copy RNN experiment. The estimator recovered the true poles 0.93 and 0.62 with maximum modulus error 0.0034, demonstrating the core fitting mechanism. However, the proposed regularization worsened task MSE (0.9456 vs 0.8289) and increased spectral radius (1.0861 vs 1.0636), so no training or stability win was observed.",
  "metrics": {
    "baseline": "Delayed-copy MSE 0.8289; spectral radius 1.0636",
    "idea": "Delayed-copy MSE 0.9456; spectral radius 1.0861; controlled pole recovery error 0.00338"
  },
  "how_to_run": "/home/maxwelhelp/main/bin/python3 experiment.py",
  "files": [
    "experiment.py",
    "results.json",
    "README.md"
  ],
  "limitations": "Only a tiny nonlinear delayed-copy task was tested; delayed-MNIST, S4/LRU/orthogonal-RNN baselines, wall-clock/FLOP matching, and multi-seed evaluation were not tested. The training penalty uses detached fitted poles to gate a spectral-norm proxy rather than backpropagating through the Hankel pole estimator, and the CUDA-failure fallback was not exercised in the successful run."
}