# Эксперимент: Universal Clock Regularization for Recurrent Dynamics (#404)

{
  "worked": true,
  "confidence": 8,
  "verdict": "Built a self-contained clock regularization MVP with a phase head, wrapped short-window phase differences, variance/cycle losses, analytical toy sweeps, and a fixed-seed GRU comparison. The core predictions manifested: phase-distortion variance and normalized clock error matched the derived scalings within 0.004% and 0.002%, noisy finite differences matched the 2σ²/Δt² law within 0.27%, and unwrap remained correct through increment 3.13 then switched branches above π as predicted. In the GRU toy, clock_E fell from 17.25 to 0.307 (98.2% lower), while one-step MSE worsened from 0.00519 to 0.00591 (13.8% higher), so this is a mechanism success and promising phase-consistency signal, not evidence of improved forecasting accuracy.",
  "metrics": {
    "baseline": "Toy GRU: one_step_mse=0.00519336, clock_E=17.2536",
    "idea": "Toy GRU: one_step_mse=0.00591047, clock_E=0.307416; analytical max relative errors: distortion variance 3.98e-5, distortion E 1.91e-5, noise variance 0.00270; unwrap boundary observed at π"
  },
  "how_to_run": "/home/maxwelhelp/main/bin/python3 clock_experiment.py",
  "files": [
    "clock_experiment.py",
    "results.json",
    "run_output.txt"
  ],
  "limitations": "Only a short synthetic sinusoid GRU was tested; no Van der Pol, Lorenz-63, long-horizon free-running forecasts, λ sweep, limit-cycle breakdown prediction, matched-parameter phase-head ablation, or multi-seed confidence intervals were run. CUDA/cuDNN failed with CUDNN_STATUS_INTERNAL_ERROR_HOST_ALLOCATION_FAILED, so the reported GRU run used the implemented CPU fallback."
}