CIMNE Seminar - "Hybrid optimization methods", by Jordi Pons
Wednesday, December 5th, 2018. Time: 12 noon
Place: O.C. Zienkiewicz Conference Room, C1 Building, UPC Campus Nord, Barcelona
Genetic algorithms are optimization methods based on the Darwin's theory of the evolution of the species. This evolution is managed by genetic information compound only by the information of individual evaluations of the function of interest. This is one of the main drawbacks when comparing with Gradient-based methods. In order to overcome this issue, hybrid methods increase the genetic information defining several populations in parallel. These populations can be defined using the same parameters, or can be different. The present presentation explore the benefits of defining population from pretty similar, just slightly modifying few parameters, to completely different, using a completely new set-up. Applications to transport simulation, or CFD optimization will be presented to illustrate the results.
Mechanical engineer in aeronautical field; developing and applying new optimization techniques, robust design optimization, hierarchical methods, game strategies, uncertainty quantification, random number generation. But also aerostructural and aeroelasticity research as main focuses of my research career.
alliances aulas awards biomechanics biomechanics and health cerca cimne CIMNE cimne-tecnologia civil-engineering coffee talk companies congress course dissemination energy-and-environment ERC-awards gid iacm ict media new building r+d+i research researchers research-overview science for society seminars structural engineering software technology-transfer transport upc velassco worldwide