Modeling issues and implementation of language for disjunctive programming

Abstract

This paper describes a number of key modeling issues for the development of tools for solving nonlinear discrete/continuous problems where logic/disjunctive constraints are included in the formulation. A generalized hybrid representation of these problems is presented. A comparison between Constraint logic programming (CLP) and generalized disjunctive programming (GDP) is established together with several constraint transformations from CLP to GDP. The components and expressions of a modeling language for setting up disjunctions and logic constraints are proposed. The language allows the specification of problems with complex logic formulations. A parser is developed for the analysis and translation of the logic sentences into files ready to be used by the solvers. An overview of the solution algorithms is also given together with several aspects about the implementation as a superset of GAMS mathematical programming language. Examples illustrating the capabilities of the proposed system are described.