Accelerated Numerical Optimization with Explicit Consideration of Model Constrains

IPARRAGUIRRE, JAVIER; DIAZ, ARIEL; BLANCO ANÍBAL M.; DAMIANI, LUCIA

Buenos Aires

Conferencia; LATIN AMERICA HIGH PERFORMANCE COMPUTING CONFERENCE; 2017

Universidad de Buenos Aires (Argentina) and Universidad de la República (Uruguay).

Population based metaheuristics can benefit from parallelization inorder to address complex numerical optimization problems. Typical realisticproblems usually involve non-linear functions, integer variables and many constraints,making the identification of optimal solutions mathematically challengingand computationally expensive. In this work, a parallelized version of theParticle Swarm Optimization technique is proposed, whose main contribution isthe explicit consideration of constraints. The implementation is tested on a classicset of optimization problems. Speedups up to 101x were obtained using asingle GPU on a standard PC using the Py-Cuda technology.