PnrResolution
Number-resolving readout with a parametrised confusion matrix
8/8 passed in 39ms
| Test | What it does | Result |
|---|---|---|
test_tes_absorb | The top outcome absorbs, so backgrounds of 0.1, 2 and 20 counts leave completeness intact and a background of 20 lands past 0.9 in that outcome. | ✅ pass |
test_tes_complete | With efficiency, background and a finite readout width the elements still sum to the identity and stay positive semi-definite. | ✅ pass |
test_tes_confuse | The readout confusion matrix is stochastic on the reported number and its interior diagonal is flat under a linear response. | ✅ pass |
test_tes_floor | The vacuum bias is upward and grows with the readout width, a count having nothing below zero to absorb into. | ✅ pass |
test_tes_identity | Unit efficiency, no background and a zero-width readout give Pi_k = |k><k| to the last bit. | ✅ pass |
test_tes_mean | The mean reported number is the mean detected number plus the readout's vacuum bias weighted by the chance the detected number was zero. | ✅ pass |
test_tes_saturate | A saturating response resolves high numbers worse than low ones, costing the top interior number more than half against a linear response. | ✅ pass |
test_tes_width | Resolution falls monotonically with the readout width and is exact at width zero. | ✅ pass |