Extended mathematical programs are collections of functions and variables joined together using specific optimization and complementarity primitives. This paper outlines a mechanism to describe such an extended mathematical program by means of annotating the existing relationships within a model to facilitate higher level structure identification. The structures, which often involve constraints on the solution sets of other models or complementarity relationships, can be exploited by modern large scale mathematical programming algorithms for efficient solution. A specific implementation of this framework is outlined that communicates structure from the GAMS modeling system to appropriate solvers in a computationally beneficial manner.Example applications are taken from chemical engineering.

The convex hull relaxation (CHR) method (Albornoz in Doctoral Dissertation, 1998, Ahlatçıoglu in Summer paper, 2007, Ahlatçıoglu and Guignard in OPIM Dept. Report, 2010 provides lower bounds and feasible solutions on convex 0-1 nonlinear programming problems with linear constraints. In the quadratic case, these bounds may often be improved by a preprocessing step that adds to the quadratic objective function terms that are equal to 0 for all 0-1 feasible solutions yet increase its continuous minimum. Prior to computing CHR bounds, one may use Plateau's quadratic convex reformulation (QCR) method ( 2006), or one of its weaker predecessors designed for unconstrained problems, the eigenvalue method of Hammer and Rubin (RAIRO 3:67-79, 1970) or the method of Billionnet and Elloumi (Math. Program, Ser. A 109:55-68, 2007). In this paper, we first describe the CHR method, and then present the QCR reformulation methods. We present computational results for convex GQAP problems.M. Guignard was partially supported under NSF Grant DMI-0400155.