-
Notifications
You must be signed in to change notification settings - Fork 77
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
is kissat non-deterministic? #49
Comments
It should be deterministic across machines, operating systems and configurations. Well there is one issue if the file has options in it and you have compiled with |
interesting, and confirmed by: then what is the meaning of "random" in the options descriptions I cited? is it "deterministic pseudo-random"? where does the seed come from? "hardware glitch" - ah. where can I buy more of these ... I am pretty sure the model was correct (my application is like a bitblasting SMT solver, and it did check the model after conversion from bits). |
|
Probably I should also add a
|
... meanwhile, I did rediscover the "lost" model. There were several moving parts (different machines, different kissat versions, changes in my bit-blaster) https://git.imn.htwk-leipzig.de/waldmann/pure-matchbox/-/blob/master/proof/Standard/Gebhardt/13.log (arctic matrix interpretation, encoding/method see https://doi.org/10.29007/qqvt , now solving an open problem. You read it here first!) |
Good, then I close this. |
The following may sound a bit strange, but ...
I think kissat (3.1.1) found a model for a CNF (roughly 200k vars, 1m clauses), using
--sat --quiet 1
, after perhaps 15 min, but I am unable to reproduce it, in several tries, giving more than 1 hour each. So - was I hallucinating? Or was kissat just lucky?The
--help
text includesmeaing that it was random?
Assuming yes: what's the best strategy to find a model (again)? Just run repeatedly, with timeout? Or run once, without timeout? (of course I will try both.)
The text was updated successfully, but these errors were encountered: