LossTolerant

Tamaki's exact phase-error estimate, and what it stops depending on

8/8 passed in 2ms

Test	What it does	Result
test_basis_flat	The single-photon detection probability does not move with the flaw or the tilt to 1e-16.	✅ pass
test_model_excess	Pereira's yield overshoots a probability by 2dark at unit efficiency against Tamaki Appendix C's 2.5dark - dark^2.	✅ pass
test_pereira_channel	flaw_channel at tilt = -delta reproduces Pereira, Curty & Tamaki's Y_Z and Eq. (33) bit error rate to a part in 1e9 wherever their form stays below 1/2.	✅ pass
test_phase_cancels	At tilt = 0 the phase error is independent of the flaw to 1e-15 across the domain.	✅ pass
test_phase_flat	The phase error is 9.920716e-04 at 0 dB against 2.979953e-03 at 40 dB, and is loss-independent at zero dark count.	✅ pass
test_phase_reads_tilt	The dark-free phase error is sin^2(tilt/2) at every flaw and transmittance, reading (1 - cos delta)/2 at the published tilt = -delta.	✅ pass
test_tamaki_ratio	The engine's dark-free sin^2(delta/2) and Tamaki Appendix C's sin^2(3*delta/8) stand in a 4/3 ratio of error angles, 9.919218567e-04 against 5.580367924e-04 at DELTA_A.	✅ pass
test_theorem_exact	Inverting the rejected data and measuring the virtual states give the same phase error rate to 1e-15 over flaw, loss, dark count and tilt.	✅ pass