PLAPIQUI   05457
PLANTA PILOTO DE INGENIERIA QUIMICA
Unidad Ejecutora - UE
artículos
Título:
Solving The Fermat-Weber Problem With a Parallel GBB Algorithm
Autor/es:
ARDENGHI JI; VAZQUEZ G.E.; BRIGNOLE N.B.
Revista:
DISCRETE APPLIED MATHEMATICS
Editorial:
ELSEVIER SCIENCE BV
Referencias:
Lugar: Amsterdam; Año: 2011
ISSN:
0166-218X
Resumen:
A location problem with fixed regions but non fixed points can be formulated
as a geometrical problem. The Global Barzilai and Borwein method (GBB)
has shown a very good performance in the resolution of this kind of problems,
finding optimal locations with small and medium size configurations. But
when a continuous region must be approximated by a discrete set of points,
the number of variables makes the computational time extremely high. In this
work we present a variation of the GBB algorithm that takes advantage of the
new multi core architecture to solve the location problem. A multi thread
algorithm based on the standard OpenMP with a parallel search and a parallel
evaluation of objective function and gradient produces results with nearly 80 %
of efficiency. Numerical results are presented.