Closed jurajsic closed 3 weeks ago
@kilohsakul says that when he was updating the code for antichain he did not touch counterexample generation, so it probably computes bullshit.
It seems the problem is in this test https://github.com/VeriFIT/mata/blob/15282c5eeaa7d786da27d0d73171679237f5c1f0/src/nfa/inclusion.cc#L163
lengths_incompatible
is some heuristic to find that the inclusion does not hold sooner and counterexample will then be a prefix of some word from the first automaton.
When I call https://github.com/VeriFIT/mata/blob/17ffaf6e2a90a506c21abe5075893b505947322b/include/mata/nfa/algorithms.hh#L77 with the following two automata:
and
I get incorrect counterexample, a word
65, 66, 66, 66, 66, 65, 67, 67
. It seems that it missed the last symbol67
.