ProtocolPhase

Phase 5b API: the phase-keyed family with its QBER simulated, not pinned

8/8 passed in 557ms

Test	What it does	Result
test_alice_plane	The rate is built on Alice's mu, not mu_bob = T*mu, and sits below the Bob-plane rate.	✅ pass
test_click_provenance	Before a run every simulated quantity is labelled still to compute, after one it is derived.	✅ pass
test_dead_time_reaches	A microsecond of dead time at a 1 GHz clock costs clicks and key, while afterpulsing adds counts and raises the QBER until no key is left.	✅ pass
test_doubles_squashed	Double clicks are kept with a random bit (squashing model): clicks = sifted + doubles, and p_click equals the sift rate.	✅ pass
test_engine_agreement	Every reported number is bit for bit what _core.run_clicks and _core.dps_rate return for the same hardware.	✅ pass
test_linewidth_costs	The QBER rises monotonically with laser linewidth and the key falls, reaching zero at 10 MHz.	✅ pass
test_one_definition	The simulated p_click matches p_ideal = 1 - (1 - dark)^2exp(-muT*eta) to 5 sigma, and dead time only lowers it.	✅ pass
test_pinned_closed_form	Pinning the error rate gives the Waks-Takesue-Yamamoto closed form (PRA 73, 012344): at e = 0 the rate is p_click*(1 - 2mu) with p_click = 1 - (1 - dark)^2exp(-muTeta), and no clicks counted.	✅ pass