Title :
A particle system for global optimization
Author_Institution :
Coordinated Sci. Lab., Univ. of Illinois at Urbana-Champaign (UIUC), Urbana, IL, USA
Abstract :
A particle algorithm for global optimization is presented, based on concepts from optimal transport. The particle flow is constructed by considering a time-stepping variational problem, and the resulting flow is interpreted as a gradient flow with respect to a certain (pseudo-) metric. The construction requires to solve an Euler-Lagrange boundary value problem at each time step, and a Galerkin procedure is described to approximate its solution. The algorithm is illustrated by a numerical example.
Keywords :
Galerkin method; boundary-value problems; optimisation; Euler-Lagrange boundary value problem; Galerkin procedure; global optimization; gradient flow; optimal transport; particle algorithm; particle flow; time-stepping variational problem; Approximation algorithms; Approximation methods; Convergence; Equations; Method of moments; Optimization; Vectors;
Conference_Titel :
Decision and Control (CDC), 2013 IEEE 52nd Annual Conference on
Conference_Location :
Firenze
Print_ISBN :
978-1-4673-5714-2
DOI :
10.1109/CDC.2013.6760129
