Enhanced global optimization methods applied to complex fisheries stock assessment models

D. Rodríguez Penas, A. Gómez, B. B. Fraguela, M. J. Martín, S. Cerviño

Statistical fisheries models are frequently used by researchers to study the behavior of marine ecosystems or to estimate the maximum acceptable catch of different species of commercial interest. The Globally applicable Area Disaggregated General Ecosystem Toolbox (Gadget) is a flexible framework that allows the development of complex statistical marine ecosystem models, being the parameters of these models usually adjusted through the use of optimization algorithms.

With the aim to improve these optimization processes, we propose an enhanced template called PMA (Parallel Multirestart Adaptive), applied to popular metaheuristic DE and PSO, adding them the following features: (1) parallelism, (2) an automatic selection of the internal settings of the algorithms, and (3) a restart mechanism to avoid local minima.

Experimental results prove that the new algorithms are faster and produce more accurate solutions than the other parallel optimization methods already included in Gadget.

Keywords: Global optimization Marine ecosystem models Particle Swarm Optimization Differential evolution

D. García Heredia, A. Alonso Ayuso, M. Laguna, E. Molina

I. Giménez Palacios, F. Parreño Torres, M. T. Alonso Martínez, R. Alvarez-Valdes