Originally posted by: GrEm

Hi
I am solving a problem with piecewise linear constraints in OPL using OPL script.
One of thoses constraints is non-convex and yields to a MIP with indicator variables.
I would like to relax all indicator constraints and solve the linear relaxation of the problem.
Is there a way to do that ?

convertAllIntVars() does not work on indicator variables/constraints apparently.

I found a reference on the the following function, but I can't make it work in OPL.
c.indicator_constraints.delete()

(found in CPLEX Python API Reference Manual)

Thanks

Hi,

do you want to do this in order to have some sensitivity analysis for a solution ?

regards

Originally posted by: GrEm

Hi

I would like to solve a relaxed version of the subproblem at some iterations, to speed up the computation.

Currently I have to re-generate the whole problem when I want to switch from the MIP formulation (with piecewise() hence with indicator constraints) to the LP relaxation.

I lose quite a lot of time in the generate() function, while only a few indicator constraints need to be relaxed.

Thanks

Hi,

would the root node relaxation help ?

(Cf https://www.ibm.com/developerworks/community/forums/html/topic?id=b334a286-b9fa-47d2-ad03-d1308bb8ff26&ps=25 )

regards