ExpHughesB92

Tier B: Hughes et al. 1999, plain B92 over 48 km of installed fibre

6/6 passed in 1ms

Test	What it does	Result
test_hughes_bounded	Their 10 Hz and 3.4 Hz are 11.6% and 5.7% of the conclusive ceiling, while their own stated factors put the first ceiling at 8.885 Hz, under the 10 they quote.	✅ pass
test_hughes_dark	The fitted background is 1.535e-4 per gate, 1.92e-5 per 730 ps window in their normalisation, a 26.3 kHz noise rate inside their 10-100 kHz.	✅ pass
test_hughes_floor	The measured visibility gives (1 - V)/2 = 0.505%, 5.43% of their 9.3% BER, and across the +/-1.24-point bar brackets the 10% their "approximately 90% of the errors are attributable to detector dark counts" leaves.	✅ pass
test_hughes_fringe	Their fringe scan's 1102 errors in 21 524 bits (4.871%) and 128-bit sample's 6 (4.688%) agree to 0.18 points, both under the 9.3% whole-set BER.	✅ pass
test_hughes_poisson	Poisson mu = 0.63 leaves 53.26% of pulses empty against their 53% and 28.21% of non-empty ones multiphoton against 28%; mu = 0.39 gives 18.24% against 18%.	✅ pass
test_hughes_predict	Declared miss: the background fitted at mu = 0.63 predicts 13.31% at mu = 0.39 against their 17.8%, and the reverse fit 12.86% at 0.63 against 9.3%.	✅ pass