[Coin-discuss] Construct a MIP problem

[Ted Ralphs]

Spencer Fung wrote:
> I see. Thanks for your response.
> 
> BTW, I just want to use the CGL to generate cuts, instead of solving the
> problem. Does it affect the cuts generation if the problem cannot be
> specified to be MIP?

I'm not sure I understand your question precisely, but compatibility 
with CGL is the main reason for the current implementation of OsiCpx. If 
you want to use CPLEX to solve an LP relaxation of a MIP and then 
subsequently load the solution into CGL, you have to avoid changing the 
problem type in CPLEX to MIP in the meantime or the solution to the LP 
relaxation will be deleted and you will have nothing to separate. But 
while the problem type is LP, you do not have access to the integrality 
status of the variables, so that information cannot be loaded into CGL. 
Something has to be cached in order for things to work properly. In 
short, the current version of OsiCpx should work properly with CGL. Does 
that answer your question?

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