> Suppose that my cost function depends on a SOS with 4 variables, i.e. > 4 regions where I decide to linearise my cost function. When I say > cost function I mean a function describing total cost of production if > production falls into that region, not unitary cost of production as I > meant in my original post.
You may look at the following note: http://winglpk.sourceforge.net/media/glpk-sos2_02.pdf which explains how to model non-convex piecewise linear functions and sos2 constraints with binary variables. _______________________________________________ Help-glpk mailing list [email protected] http://lists.gnu.org/mailman/listinfo/help-glpk
