You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
$ cvc5 -q unknown.smt2
unknown
$ cvc5 -q --sygus-inst unknown.smt2
sat
$ cvc4-1.8 -q unknown.smt2
sat
$ time cvc5 --force-logic=NRA unknown.smt2
^Ccvc5 interrupted by user.
real 1m7.944s
user 1m7.907s
sys 0m0.020s
$ cat unknown.smt2
(declare-const x Real)
(declare-const u Real)
(declare-const v Real)
(assert (exists ((t Real)) (and (>= x (- 1.0)) (not (or (< t 0.0) (= t (* u t v)))))))
(check-sat)
The text was updated successfully, but these errors were encountered:
Fixescvc5/cvc5-projects#215Fixescvc5/cvc5-projects#291Fixescvc5/cvc5-projects#292Fixescvc5/cvc5-projects#294Fixescvc5/cvc5-projects#297
Previously the concern was that this would interfere with e.g. quantifiers + non-linear. However, our strategy is now fair wrt other theories as we send lemmas at LAST_CALL effort, and is properly restricted by the nl-ext mode.
Moreover, this option increases our ability to solve problems (instead of saying "unknown") significantly, even for quantified logics like UFNIA.
Commit: 1639655ca7b8f0f18145fdbb515253810b119d08
Another small NRA regression, decidable by
--sygus-inst
but not by forcing the logic. Possibly related to https://github.com/cvc5/cvc5/issues/6798.The text was updated successfully, but these errors were encountered: