ExpDiamantiJitter

Tier B: Diamanti et al. 2006, the 3.4% QBER with timing jitter modelled

7/7 passed in 169ms

Test	What it does	Result
test_gaussian_excluded	A pure Gaussian fitted to the 100 ps loss needs 192.1 ps FWHM, predicts 22.0% loss at 200 ps against 40%, leaks zero and gives 2.691% QBER instead of 3.4%.	✅ pass
test_near_gap	Declared miss: the response fitted at 100 km predicts 2.795% QBER at 10 km against their 2.2%, while the jitter-free reading sits below 2.2%.	✅ pass
test_qber_total	The response fit reproduces 3.4% and the 0.7% jitter share by construction, and predicts the rest splits 1.726% dark against 1.7% and 0.966% interferometric against 1%.	✅ pass
test_rate_unmoved	Deriving the 54% window cost leaves the 100 km sift at 2.02816e-6 and the 166 bit/s anchor at 168.19.	✅ pass
test_response_plausible	The 79.33 ps core leaves 44.0 ps prompt jitter beside the 66 ps pulse, the 295 ps tail sits in the 0.2-3 ns diffusion band, and a 0.1-point QBER shift moves the core 9.1 ps.	✅ pass
test_timed_montecarlo	At 10 km, 4e6 pulses of run_clicks with the response on reproduce the sift to 0.2% and QBER to 0.11 points, at 40.000% window loss and 2.33% bin leak.	✅ pass
test_window_costs	One fitted response gives their 54% loss at 100 ps and 40% at 200 ps, 1.30x more light from doubling the window where a single Gaussian gives 1.44x.	✅ pass