Open jonnybest opened 11 years ago
Maybe I could fix this by introducing a lemma for each call containing a2r as an argument. The question remains: How do I implement this efficiently? So far I can only write lemmas for each declaration. I would have to extend the program to allow for for this more 'dynamic' approach.
The a2r axiom presents a difficult problem for MBQI. This problem can be solved by MBQI only while adding this axiom:
This lets the MBQI-enabled solver fail. The problem for MBQI is the equality between the
in_1
part and theforall
part. The solver behaves the same way for substitutingand
with=
or even with"<=>"
as two assertions.You can observe that MBQI succeeds if you drop ax5_2 from the VC.
I have gotten the idea of testing MBQI only from this thread on the stackoverflow http://stackoverflow.com/questions/13025127/getting-sat-for-large-problems/13032820