No abstract

No abstract

To properly describe and solve complex decision problems, research on theoretical properties and solution of mixed-integer quadratic programs is becoming very important. We establish in this paper different Lipschitz-type continuity results about the optimal value function and optimal solutions of mixed-integer parametric quadratic programs with parameters in the linear part of the objective function and in the right-hand sides of the linear constraints. The obtained results extend some existing results for continuous quadratic programs, and, more importantly, lay the foundation for further theoretical study and corresponding algorithm analysis on mixed-integer quadratic programs. §1 Introduction Quadratic programs (QPs) can be used to model and solve many management problems such as investment analysis ([17], [8]). They have been playing a very important role in the theoretical study and solution algorithm design of nonlinear programs ([18], [22], [21]). Nevertheless, in order to extend the modeling ability of usual continuous QPs and to properly describe complex decision problems involving logical or indivisible constraints, disjunctions or piecewise linearity, integer variables have to be introduced. Typical situations include control and communication problems [1], and the economic dispatch of generator with prohibited operating zones [19]. Researches on (mixed-)integer nonlinear programming problems, especially mixed-integer quadratic programs (MIQPs) have attracted more and more attention during the last decade. Several useful algorithms have been designed to solve MIQPs, for instance, the branch-and-cut algorithm [7], the Lagrangian-based decomposition and linearization technique for 0-1 quadratic programming [10], the general-purpose MIQP solver of the ILOG software [13]. To investigate the convergence and robustness of these algorithms one needs to know the vari-