DICOPT

DICOPT solves mixed-integer nonlinear programming (MINLP) model types.  It was developed by J. Viswanathan and I.  Grossmann at Carnegie Mellon University.  The models solved involve linear binary or integer variables and linear and nonlinear continuous variables.  Although the algorithm has provisions to handle non-convexities, it does not necessarily obtain the global optimum.

DICOPT implements extensions of the outer-approximation algorithm for the equality relaxation strategy.  The DICOPT solution approach involves solving a series of NLP and MIP sub-problems.  These sub-problems can be solved using any NLP or MIP solver that runs under GAMS.  This allows one to match the best algorithms to the problem at hand and guarantees that enhancements in the NLP and MIP solvers are exploited.  It also requires licenses to such solvers.  DICOPT solves MINLP model types.  The solver manual is on DICOPT.pdf.