ept.git/ep_run/r_sweep.log, branch xiang Unnamed repository; edit this file 'description' to name the repository. r-sweep: cos(EP,BPTT) FLAT across r=0.02..0.4 on both operators — nudge-SNR hypothesis refuted 2026-07-05T10:12:34+00:00 Yuren Hao yurenh2@illinois.edu 2026-07-05T10:12:34+00:00 5975caf0f52355276aabec5ba8f144db9ac46619 The 50% single-shot sensitivity is MULTIPLICATIVE (near-marginal T2 dynamics' state-sensitivity; signal and error scale with r together) — not additive noise divided by 2r. cos ceiling ~0.88 at near-edge states is set by T2 truncation + state sensitivity (0.98 at deeply-contracted states). hr-0.2's empirical wins are NOT estimator SNR. Hardware upside: algorithm indifferent to r across 20x -> nudge amplitude free to fight ADDITIVE readout noise. Surviving accuracy levers: kappa nbrake (Tikhonov, targets the real culprit), tail-window averaging over argmin selection. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_014FAPDWQ49M5Ye3NpTndTpn 
The 50% single-shot sensitivity is MULTIPLICATIVE (near-marginal T2 dynamics'
state-sensitivity; signal and error scale with r together) — not additive
noise divided by 2r. cos ceiling ~0.88 at near-edge states is set by T2
truncation + state sensitivity (0.98 at deeply-contracted states). hr-0.2's
empirical wins are NOT estimator SNR. Hardware upside: algorithm indifferent
to r across 20x -> nudge amplitude free to fight ADDITIVE readout noise.
Surviving accuracy levers: kappa nbrake (Tikhonov, targets the real culprit),
tail-window averaging over argmin selection.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_014FAPDWQ49M5Ye3NpTndTpn