[Coin-discuss] osi cplex implementation not MIP friendly

[Ted Ralphs]

Jevan,

I was in a hurry yesterday and didn't read your mail closely enough to 
understand the true nature of your problem before I replied. Upon further 
inspection, the real issue is that the current version of OSI does not really 
have much support for MIP, other than the branchAndBound() method. This will be 
fixed in the new version. I fixed the bugs you pointed out in switchToLP() and 
switchToMIP(). The problem with isProvenOptimal(), isProvenPrimalInfeasible(), 
and isProvenDualInfeasible() is that they are really only intended to process 
return codes for LP. Currently, they don't check the return codes for MIP. I 
changed isProvenOptimal() so that it should properly detect whether the MIP has 
been solved to optimality or not, but the other functions really only work for 
LP. Are these fixes good enough for now?

Ted
-- 
Dr. Ted Ralphs
Assistant Professor
Industrial and Systems Engineering
Lehigh University
(610)758-4784
tkralphs at lehigh.edu
www.lehigh.edu/~tkr2