Particle Swarm Optimization Research Toolbox
24 Jul 2010
15 May 2011)
Gbest PSO, Lbest PSO, RegPSO, GCPSO, MPSO, OPSO, Cauchy mutation, and hybrid combinations
function [f] = ObjFun_Rosenbrock(position_matrix, num_particles_2_evaluate)
% Copyright 2008, 2009, 2010, 2011 George I. Evers
%The "position_matrix" of positions to be evaluated is passed in.
%This is generally equal to the full position matrix, "x," though it
%may be a subset - such as one particular row of "x."
%Order: "num_particles_2_evaluate" x "dim" (usually "np" by "dim")
%The number of particles/rows to be evaluated, "num_particles_2_evaluate,"
%is passed in: this is generally equal to the number of particles
%in the swarm, "np." Passing this value into the function
%eliminates the iterative need to calculate size(position_matrix, 1).
% Global Variable:
%The problem dimensionality, "dim," is declared global since it does
%not change when the objective function is called. This eliminates
%the iterative need to calculate size(position_matrix, 2).
%Column vector "f" contains one function value per particle/row
%Order: "num_particles_2_evaluate" x 1 (usually "np" by 1)
% Global minimizer: ones(1, dim)
% Global minimum: f(ones(1, dim)) = 0
% Initialization space: Assuming symmetric initialization, positions
%were initialized to lie within [-30, 30] when the
%position matrix was created. This can be changed within
f = zeros(num_particles_2_evaluate, 1);
for Internal_j = 1:(dim-1)
f = (1 - position_matrix(1:num_particles_2_evaluate,Internal_j)).^2 + 100*(position_matrix(1:num_particles_2_evaluate,Internal_j+1) - (position_matrix(1:num_particles_2_evaluate,Internal_j)).^2).^2 + f;
%verified by comparing values with those generated by:
%(2) updating the code available at