Full Closure Hall And Critical ILP Do Not Refute F3 Operation Extension¶
Claim/Theorem¶
The full rank-order closure diagnostics do not refute operation extension for the current f_3 certificate.
Using the exact pattern-upset closure from [[full-rank-order-closure-reduces-to-pattern-upset-lattice.md]], the script docs/project_QEM-QEC/tmp/scripts/route_d_full_rank_order_extension.py scanned every one of the 2,849,631 domain closure elements for the interval/Hall obstruction forced by the fixed generator values.
For f_3:
- interval violations found:
0; - forced-propagation contradictions found:
0; - minimum interval slack:
0; - zero lower-slack elements:
65; - zero upper-slack elements:
65; - forced-one antichain total:
75; - forced-zero antichain total:
75.
The induced critical-subposet problem has 123 closure elements. Both the binary ILP and its fractional LP relaxation solve optimally:
- binary ILP variables:
6150; - binary ILP monotonicity constraints:
222500; - binary ILP status:
Optimal; - fractional LP status:
Optimal.
This is a genuine strengthening beyond pairwise closure: it uses all closure elements for the interval/Hall scan and all zero-slack forced elements for the critical ILP.
The controls calibrate the test:
- the toy submodularity control has closure size
4and solves optimally; - the six-qubit representable control has domain closure size
35,664, but its fan certificate fails the forced-propagation test and the critical binary/fractional programs are infeasible.
The six-qubit failure is not a contradiction. It means the six-qubit fan certificate does not lift to this operation-level invariant, which is exactly what [[fan-cone-certificates-are-not-hidden-vertex-invariants.md]] predicts for a known hidden-vertex-representable function.
Therefore the current full-closure diagnostics are nontrivial and correctly reject at least one cone-specific certificate, but they still do not decide the f_3 certificate.
Dependencies¶
- [[full-rank-order-closure-reduces-to-pattern-upset-lattice.md]]
- [[f3-certificate-passes-pairwise-rank-order-extension-but-not-full-closure.md]]
- [[fan-cone-certificates-are-not-hidden-vertex-invariants.md]]
- [[six-qubit-witness-is-hidden-vertex-graph-cut-representable.md]]
Conflicts/Gaps¶
- The critical-subposet ILP is a necessary restriction, not the full monotone label-function problem.
- No operation map is extracted for
f_3. - No infeasibility certificate is found for
f_3. - The result narrows the obstruction side to a larger monotone count-decomposition problem rather than closing it.
Sources¶
10.1016/j.dam.2009.07.001- local computation:
docs/project_QEM-QEC/tmp/certificates/full_rank_order_extension_report.json - local script:
docs/project_QEM-QEC/tmp/scripts/route_d_full_rank_order_extension.py