Groupe d’études et de recherche en analyse des décisions

Random derivative-free algorithm for continuously differentiable nonlinear problems

Majid Jaberipour Université de la Colombie-Britannique, Canada

Derivative-Free optimization is an area of long history which has so many applications in different fields. This paper describes a random derivative-free algorithm for solving unconstrained or bound constrained continuously differentiable nonlinear problems. This method is a combination of Particle swarm and directional direct search algorithm. At first glance, a simple way of generating positive bases has been introduced for solving continuously differentiable problems. Then, it has been shown that using Particle swarm algorithm with direct search algorithm can solve nonlinear optimization problems with high dimensions efficiently.