U. Aldasoro, G. Pérez-Valdés, A. Pagès-Bernaus, A. Tomasgard
Branch-and-Fix Coordination is an algorithm for solving large scale mixed integer (binary) optimization problems, in which the original model is decomposed into submodels using break-stage scenario clustering. Thus, we consider a master-worker distributed parallel computing implementation: worker processors solve the cluster submodels in parallel, whereas the master processor coordinates the values of the integer variables. In this work, we present several nested decomposition schemes for solving the full size LP problem that guarantees non-anticipativity in the continuous variables of the BFC master problem. This idea allows solving problems of a larger scale, as well as a more efficient use of the parallel computing resources.
Keywords: Stochastic optimization, decomposition methods, distributed computing, large scale optimization
Scheduled
WA1 Decomposition methods for Stochastic Programming
June 1, 2016 9:00 AM
Salón de actos
