Skip to content

ExpRubenokRelay ​

Tier B: Rubenok et al. 2013, MDI-QKD measured over deployed fibre

8/8 passed in 8ms

TestWhat it doesResult
test_mdi_bellReading Q_vv as a projection onto psi- alone gives p_d = 1.884e-5, 3.0% from their companion paper's 1.83e-5, against 27% for Ma and Razavi's four-pattern reading.✅ pass
test_mdi_decoyXu, Curty, Qi and Lo's single-photon-pair bound on their 3x3 diagonal grid sits within 2.1% of the halved forward model at all four setups, above it on the error rate, and ignores the two unpublished cells.✅ pass
test_mdi_deficitDeclared miss: the rectilinear QBER is exact at fitted setup 2 but 0.04429 against a measured 0.053 at 18.2 dB, a channel background the Ma-Razavi model has no term for.✅ pass
test_mdi_efficiencyEach gain inverted for detector efficiency alone gives a 10.2% spread straddling their companion paper's 0.145.✅ pass
test_mdi_gainsAll sixteen measured gains across four setups, two bases and two intensities sit within 0.876x to 1.051x of the model, nothing fitted.✅ pass
test_mdi_qberOne misalignment fitted to setup 2's rectilinear QBER puts all four diagonal QBERs within 0.8 points of measurement, near Ma and Razavi's 1/4+ed/2.✅ pass
test_mdi_ratesRebuilt from their grid the key rates are 2.94x to 7.36x their published figures, bounded not reproduced, their overlaps being "insufficient to securely distribute key".✅ pass
test_mdi_visibilityThe Poissonian HOM ceiling is exactly their "maximum value of 50%", their 47±1% inverts to 0.940 mode overlap, and 500 ps pulses at 30 ps drift account for 0.50 of its 6.0 missing points.✅ pass