New Formulations and Branching Strategies for the GOP Algorithm

摘要: In Floudas and Visweswaran (1990, 1993), a deterministic global optimization approach was proposed for solving certain classes of nonconvex problems. A algorithm, GOP, presented the solution problem through series primal relaxed dual problems that provide valid upper lower bounds respectively on solution. The algorithm proven to have finite convergence an r-global optimum. this paper, branch-and-bound framework GOP is presented, along with several reduction tests can be applied at each node tree. effect properties prune tree tighter underestimators We also present mixed-integer linear programming (MILP) formulation problem, which enables implicit enumeration nodes in iteration. Finally, alternate branching scheme number subproblems. Simple examples are illustrate new approaches. Detailed computational results implementation both versions found companion paper chapter 4.