Add HiGHS.ComputeInfeasibilityCertificate attribute #274
Conversation
Codecov ReportAll modified and coverable lines are covered by tests ✅
Additional details and impacted files@@ Coverage Diff @@
## master #274 +/- ##
=========================================
Coverage 100.00% 100.00%
=========================================
Files 1 1
Lines 1471 1479 +8
=========================================
+ Hits 1471 1479 +8 ☔ View full report in Codecov by Sentry. 🚀 New features to boost your workflow:
|
|
cc @sstroemer thoughts? |
sstroemer
left a comment
sstroemer
left a comment
There was a problem hiding this comment.
Noticed two minor things, but looks good to me - thanks for adding that specifically!!
src/MOI_wrapper.jl
Outdated
| x.model_status = Highs_getModelStatus(model) | ||
| statusP = Ref{HighsInt}() | ||
| if x.model_status == kHighsModelStatusInfeasible | ||
| certificates = MOI.get(model, ComputeInfeasibilityCertificates()) |
There was a problem hiding this comment.
One thing I noticed: Setting ComputeInfeasibilityCertificates() to false will always result in both x.has_dual_ray and x.has_primal_ray being false.
Therefore that will never trigger:
if x.has_dual_ray || x.has_primal_ray
return # If a ray is present, we don't query the solution
endI have no idea if querying the solution (status) afterwards costs time or whether skipping it also when !certificates && (x.model_status == kHighsModelStatusInfeasible || x.model_status == kHighsModelStatusUnbounded) would even be correct, but just a thought.
There was a problem hiding this comment.
We query the solution only if x.primal_solution_status != kHighsSolutionStatusNone, which won't be the case for your infeasible problems. I'd like to leave the behavior as-is.
odow
changed the title
Closes #273