MdiProgramme
The nine-cell decoy bound as a linear programme, against the seven-cell form
11/11 passed in 33090ms
| Test | What it does | Result |
|---|---|---|
test_cut_domain | A truncation below 2 (no (1,1) term) or above 16 (past the solver's row cap) is refused. | ✅ pass |
test_empty_cell | An empty grid cell, leaving no strictly feasible primal point, is refused. | ✅ pass |
test_error_sandwich | At five spans the error programme sits above the forward e11 and below the four-cell closed form. | ✅ pass |
test_exact_vacuum | Against a true vacuum the four-cell e11 is already exact and the error programme refuses. | ✅ pass |
test_pinned_gain | At Curty's intensities and 100 km the nine cells tighten Y11 by 3.6% and e11 by 1.9% over the seven. | ✅ pass |
test_random_sandwich | Over twenty random spans, ladders and off-centre relays every accepted certificate stays under the forward yield, and more than fourteen certify. | ✅ pass |
test_rate_improves | At 0 and 100 km the programmes raise the rate by 16% at Curty's intensities and by under 2% against an exact vacuum. | ✅ pass |
test_small_cut | Cutoff 4 at 200 km, valid but no tighter than the closed form, is refused. | ✅ pass |
test_vacuum_decides | At 50 and 150 km the programme tightens Y11 by under 1% at an exact-zero vacuum and by over 3% at 5e-4. | ✅ pass |
test_yield_bracket | At five spans the programme's Y11 sits under the forward yield and above the closed form returned beside it, with a positive central-path gap and non-zero iterations. | ✅ pass |
test_yield_needed | The error programme refuses y11 = 0. | ✅ pass |