Closed geoffgeoffgeoff3 closed 1 year ago
The problem here is that in PUZ001-1.p, the negated conjecture is not of the form "not something", but the -olpterm option expects it to be of that form because normally it is called from problems with a conjecture. Do you need the -olpterm option for this problem? I am not completely sure how to do in the case the negated conjecture is not a negation.
Hi Guillaume,
The problem here is that in PUZ001-1.p, the negated conjecture is not of the form "not something", but the -olpterm option expects it to be of that form because normally it is called from problems with a conjecture.
What?!!? How do you guys ever prove anything like ... ~ sensible(donald_trump) ? That seems to be a fundamental flaw in lambdapi (or something).
Do you need the -olpterm option for this problem? I am not completely sure how to do in the case the negated conjecture is not a negation.
Vote Republican, I guess :-?
Cheers,
Geoff
"My cat" is not a float. Every string should learn to swim. Despite the high cost of living, it's still surprisingly popular. Does superman wonder why we wear our overpants on the inside?
Fortunately, Zenon Modulo can prove
~ sensible(donald_trump)
because it will show that the negated conjecture
~ ~ sensible(donald_trump)
is inconsistent.
In fact, Zenon Modulo can prove negated conjectures even if they do not begin with ~. It can even produce a Lambdapi proof in that case, with the option -olp (although I didn't check that it works well).
The problem comes from the -olpterm option which assumes that one tries to prove a conjecture, so that it outputs a proof term for that conjecture.
One could probably output something also in the case we have a negated conjecture that is not a negation, but I am not sure where it could be useful. To my knowledge, -olpterm is only used in Ekstrakto and GDV, to prove conjectures.
In which context did you tried to use it? Is it within GDV?
Hi Guillaume,
Fortunately, Zenon Modulo can prove
~ sensible(donald_trump)
because it will show that the negated conjecture~ ~ sensible(donald_trump)
is inconsistent.
:-)
In fact, Zenon Modulo can prove negated conjectures even if they do not begin with ~. It can even produce a Lambdapi proof in that case, with the option -olp (although I didn't check that it works well).
The problem comes from the -olpterm option which assumes that one tries to prove a conjecture, so that it outputs a proof term for that conjecture.
Yeah. My default configuration uses -olpterm ... zenon_modulo -p0 -itptp -olpterm -x arith $NNPPArg $LeafArg -max-time $1s -max-size 12G -sig LAMBDAPI_CONTEXT.Signature $PROBLEMFILE > $SOLUTIONFILE
One could probably output something also in the case we have a negated conjecture that is not a negation, but I am not sure where it could be useful. To my knowledge, -olpterm is only used in Ekstrakto and GDV, to prove conjectures.
Even in the GDV context it can be possible that the "negated conjecture that is not a negation". The negation or lack thereof should not hamper the proof processes.
In which context did you tried to use it? Is it within GDV?
I was just trying to prove ... https://www.tptp.org/cgi-bin/SeeTPTP?Category=Problems&Domain=PUZ&File=PUZ001-1.p
Cheers,
Geoff
"My cat" is not a float. Every string should learn to swim. Despite the high cost of living, it's still surprisingly popular. Does superman wonder why we wear our overpants on the inside?
it crashes on CNF problems, e.g., ... https://www.tptp.org/cgi-bin/SeeTPTP?Category=Problems&Domain=PUZ&File=PUZ001-1.p
My command line is ... zenon_modulo -p0 -itptp -olpterm -x arith -max-time 30s -max-size 12G -sig LAMBDAPI_CONTEXT.Signature PUZ001-1.p ... and the barf is ... Fatal error: exception File "lltolp.ml http://lltolp.ml", line 1060, characters 11-17: Assertion failed Raised at file "list.ml http://list.ml", line 224, characters 10-25 Called from file "expr.ml http://expr.ml", line 249, characters 45-90