geometry-certificate-rhombus-diagonals-perpendicular
- Kind
- witness-replay
- Status
- checked
Supports: the saturated cofactor certificate, re-checked from the committed artifact by a checker that shares no code with the search that produced it
cargo test -p axeyum-cas --test geometry_certificate_artifacts Evidence notes
The certificate uses the Rabinowitsch trick: a fresh variable Zinv0 and the generator `collinear(A,B,D) * Zinv0 - 1`. Specialising Zinv0 := 1/collinear(A,B,D) is legitimate exactly when A, B, D are not collinear, and that is the ONLY place the side condition enters. Three tamper controls in the same suite bear directly on this fact and every one of them now runs against EVERY saturated certificate rather than only the alphabetically first: DELETING the degenerate counterexample is rejected; REPLACING it with a generic configuration is rejected; and replacing it with A = (0,0), B = (1,0), C = (2,0), D = (1,0) -- which genuinely violates `abd-not-collinear` but does NOT falsify the conclusion -- is also rejected. That last one is the control that distinguishes a counterexample from a configuration that merely sits on the degeneracy locus.