GLOBAL OPTIMIZATION OF NONCONVEX NONLINEAR PROGRAMS VIA INTERVAL-ANALYSIS

A new global-optimization procedure is devised to tackle nonconvex non linear programming problems. The proposed algorithm is based on interv al analysis and is guaranteed to yield the global solution. Several ne w accelerating tools are introduced to significantly reduce the comput ational intensity associated with classical interval-based optimizatio n techniques. These tools include a lower-bound test as well as a ''di strust-region'' method for deleting infeasible subspaces. In addition, a local search is employed to continuously update an upper bound on t he solution and to provide a means of splitting the search space. Illu strative examples are solved to demonstrate the efficacy of this novel algorithm.