Closed scungao closed 9 years ago
I will fix it tonight. I just arrived at Pittsburgh by the way.
On Tuesday, January 20, 2015, Sicun Gao notifications@github.com wrote:
The new version hangs on the following formula. If given reasonable bounds on x then there's no problems. Something's wrong when handling larger domains.
(set-logic QF_NRA) (declare-fun x () Real) (assert (= x (+ x 10.0))) (check-sat) (exit)
— Reply to this email directly or view it on GitHub https://github.com/dreal/dreal/issues/67.
Soonho
Great. If I give big bounds it hangs as well, say
(set-logic QF_NRA) (declare-fun x () Real) (assert (< x 1000000000)) (assert (> x -1000000000)) (assert (= x (+ x 10.0))) (check-sat) (exit)
Now I suspect if this is an inherent problem and the solver really start branching on things, due to no clever preprocessing on the linear equality.
If I remember correctly Fedor observed a similar thing on the 'building planning' problems of his MSc project. (All linear stuff ...)
@scungao, your guess is right. I've added CtcPolytopeHull
ibex contractor which uses a linear solver. It reduces the running time to 1.3 sec and it gives UNSAT for your latter example.
Fantastic!
On Jan 21, 2015, at 04:10, Soonho Kong notifications@github.com wrote:
@scungao, your guess is right. I've added CtcPolytopeHull ibex contractor which uses a linear solver. It reduces the running time to 1.3 sec and it gives UNSAT for your latter example.
— Reply to this email directly or view it on GitHub.
Is this fixed in the new nra version or the main repo?
On Tue, Jan 27, 2015 at 3:28 AM, Soonho Kong notifications@github.com wrote:
Closed #67 https://github.com/dreal/dreal/issues/67.
— Reply to this email directly or view it on GitHub https://github.com/dreal/dreal/issues/67#event-225190877.
Not yet but we have a separate issue there.
On Tuesday, January 27, 2015, Sicun Gao notifications@github.com wrote:
Is this fixed in the new nra version or the main repo?
On Tue, Jan 27, 2015 at 3:28 AM, Soonho Kong <notifications@github.com javascript:_e(%7B%7D,'cvml','notifications@github.com');> wrote:
Closed #67 https://github.com/dreal/dreal/issues/67.
— Reply to this email directly or view it on GitHub https://github.com/dreal/dreal/issues/67#event-225190877.
— Reply to this email directly or view it on GitHub https://github.com/dreal/dreal/issues/67#issuecomment-71653047.
Soonho Kong soonhok@cs.cmu.edu
The new version hangs on the following formula. If given reasonable bounds on x then there's no problems. Something's wrong when handling larger domains.
(set-logic QF_NRA) (declare-fun x () Real) (assert (= x (+ x 10.0))) (check-sat) (exit)