A Simple Heuristic Optimization in Python

SHO.py

import numpy as np
import random

def sphere(x):
    tmp = 0
    for i,d in enumerate(x):
        tmp += (d - 0.1*i)*(d - 0.1*i);
    return tmp

class swarm(object):
    p = []
    r = []
    """docstring for swarm"""
    def __init__(self, num, dim, lb, ub):
        super(swarm, self).__init__()
        self.num = num
        self.dim = dim
        self.lb = lb
        self.ub = ub

        self.p = np.random.uniform(self.lb,self.ub,(self.num,self.dim))
        for row in range(0,self.p.shape[0]):
            self.r.append(sphere(self.p[row]))

    def update(self):
        # 2.1 select random particles
        r0, r1 = random.sample(range(0,self.num),2)
        rj = random.sample(range(0,self.dim),2)        # random dimension
        # 2.2 generate new_sol particle from sol via slice
        new_sol = np.copy(self.p[r0])
        new_sol[rj] = self.p[r0][rj] + np.random.uniform(-1, 1)*(self.p[r0][rj] - self.p[r1][rj])
        #new_sol[rj] = self.p[self.r.index(min(self.r))][rj] + np.random.uniform(-1, 1)*(self.p[r0][rj] - self.p[r1][rj])
        # 2.3 greedy select
        if sphere(new_sol) < sphere(self.p[r0]):
            self.p[r0][rj] = new_sol[rj]
            self.r[r0] = sphere(new_sol)

    def gbest(self):
        return min(self.r)
    def show(self):
        print('sphere',self.p[self.r.index(min(self.r))],' =', min(self.r))


def main():
    s = swarm(5,5,-100,100)
    for x in range(0,6000):
        s.update()
    s.show()

if __name__ == '__main__':
    main()