Parallel Interval Continuous Global Optimization Algorithms

We theorically study, on a distributed memory architecture, the parallelization of Hansen's algorithm for the continuous global optimization with inequality constraints, using interval arithmetic. We propose a parallel algorithm based on a dynamic redistribution of the working list among the processors. On the other hand, we exploit the reduction technique, developped by Hansen, for computing a first global upper bound which is needed for both sequential and parallel continuous global optimization algorithms.


unpublished (new paper)