Эксперимент: Periodic CMV Unitary Recurrent Layer (#1352)

{ "worked": false, "confidence": 9, "verdict": "Built a differentiable periodic CMV-style recurrent transition using alternating local real 2x2 rotation cores with period tying. The numerical check confirmed orthogonality and norm preservation to about 3.5e-6 error, while a 1.05-scaled dense orthogonal control amplified norms by 131.5x after 100 steps. However, on the toy sum-of-sequence task the CMV models reached 0.678 accuracy versus 0.775 for the dense orthogonal RNN and were about 10x slower in this unoptimized local-scatter implementation, so the claimed practical learning win was not observed.", "metrics": { "baseline": "Dense orthogonal RNN: 1,121 parameters, final loss 0.5885, accuracy 0.7754, 4.62 s; expansive dense norm ratio after 100 steps: 131.50x", "idea": "CMV period-8: 105 parameters, final loss 0.6252, accuracy 0.6777, 48.23 s; period-32: 129 parameters, final loss 0.6311, accuracy 0.6777, orthogonality error 2.38e-7 and norm-ratio maximum deviation 3.46e-6" }, "how_to_run": "/home/maxwelhelp/main/bin/python3 run_experiment.py", "files": [ "run_experiment.py", "results.json" ], "limitations": "Only a small width-32, length-40 synthetic sum task was tested; no sequential-MNIST, copy-memory, complex cores, turnover refactorization, gradient/Jacobian statistics, optimized batched factor kernels, or broader random-seed evaluation was performed." }