#!/usr/bin/env python
# Created by "Thieu" at 17:48, 21/05/2022 ----------%
# Email: nguyenthieu2102@gmail.com %
# Github: https://github.com/thieu1995 %
# --------------------------------------------------%
import numpy as np
from mealpy.optimizer import Optimizer
from mealpy.utils.agent import Agent
[docs]class OriginalESOA(Optimizer):
"""
The original version of: Egret Swarm Optimization Algorithm (ESOA)
Links:
1. https://www.mathworks.com/matlabcentral/fileexchange/115595-egret-swarm-optimization-algorithm-esoa
2. https://www.mdpi.com/2313-7673/7/4/144
Examples
~~~~~~~~
>>> import numpy as np
>>> from mealpy import FloatVar, ESOA
>>>
>>> def objective_function(solution):
>>> return np.sum(solution**2)
>>>
>>> problem_dict = {
>>> "bounds": FloatVar(n_vars=30, lb=(-10.,) * 30, ub=(10.,) * 30, name="delta"),
>>> "minmax": "min",
>>> "obj_func": objective_function
>>> }
>>>
>>> model = ESOA.OriginalESOA(epoch=1000, pop_size=50)
>>> g_best = model.solve(problem_dict)
>>> print(f"Solution: {g_best.solution}, Fitness: {g_best.target.fitness}")
>>> print(f"Solution: {model.g_best.solution}, Fitness: {model.g_best.target.fitness}")
References
~~~~~~~~~~
[1] Chen, Z., Francis, A., Li, S., Liao, B., Xiao, D., Ha, T. T., ... & Cao, X. (2022). Egret Swarm Optimization Algorithm:
An Evolutionary Computation Approach for Model Free Optimization. Biomimetics, 7(4), 144.
"""
def __init__(self, epoch=10000, pop_size=100, **kwargs):
super().__init__(**kwargs)
self.epoch = self.validator.check_int("epoch", epoch, [1, 100000])
self.pop_size = self.validator.check_int("pop_size", pop_size, [5, 10000])
self.set_parameters(["epoch", "pop_size"])
self.is_parallelizable = False
self.sort_flag = False
[docs] def generate_empty_agent(self, solution: np.ndarray = None) -> Agent:
if solution is None:
solution = self.problem.generate_solution(encoded=True)
weights = self.generator.uniform(-1., 1., self.problem.n_dims)
m = np.zeros(self.problem.n_dims)
v = np.zeros(self.problem.n_dims)
return Agent(solution=solution, weights=weights, local_solution=solution.copy(), m=m, v=v)
[docs] def generate_agent(self, solution: np.ndarray = None) -> Agent:
"""
ID_WEI = 2
ID_LOC_X = 3
ID_LOC_Y = 4
ID_G = 5
ID_M = 6
ID_V = 7
"""
agent = self.generate_empty_agent(solution)
agent.target = self.get_target(agent.solution)
agent.local_target = agent.target.copy()
agent.g = (np.sum(agent.weights * agent.solution) - agent.target.fitness) * agent.solution
return agent
[docs] def initialize_variables(self):
self.beta1 = 0.9
self.beta2 = 0.99
[docs] def evolve(self, epoch):
"""
The main operations (equations) of algorithm. Inherit from Optimizer class
Args:
epoch (int): The current iteration
"""
hop = self.problem.ub - self.problem.lb
for idx in range(0, self.pop_size):
# Individual Direction
p_d = self.pop[idx].local_solution - self.pop[idx].solution
p_d = p_d * (self.pop[idx].local_target.fitness - self.pop[idx].target.fitness)
p_d = p_d / ((np.sum(p_d) + self.EPSILON)**2)
d_p = p_d + self.pop[idx].g
# Group Direction
c_d = self.g_best.solution - self.pop[idx].solution
c_d = c_d * (self.g_best.target.fitness - self.pop[idx].target.fitness)
c_d = c_d / ((np.sum(c_d) + self.EPSILON)**2)
d_g = c_d + self.g_best.g
# Gradient Estimation
r1 = self.generator.random(self.problem.n_dims)
r2 = self.generator.random(self.problem.n_dims)
g = (1 - r1 - r2) * self.pop[idx].g + r1 * d_p + r2 * d_g
g = g / (np.sum(g) + self.EPSILON)
self.pop[idx].m = self.beta1 * self.pop[idx].m + (1 - self.beta1) * g
self.pop[idx].v = self.beta2 * self.pop[idx].v + (1 - self.beta2) * g**2
self.pop[idx].weights -= self.pop[idx].m / (np.sqrt(self.pop[idx].v) + self.EPSILON)
# Advice Forward
x_0 = self.pop[idx].solution + np.exp(-1.0 / (0.1 * self.epoch)) * 0.1 * hop * g
x_0 = self.correct_solution(x_0)
y_0 = self.get_target(x_0)
# Random Search
r3 = self.generator.uniform(-np.pi/2, np.pi/2, self.problem.n_dims)
x_n = self.pop[idx].solution + np.tan(r3) * hop / epoch * 0.5
x_n = self.correct_solution(x_n)
y_n = self.get_target(x_n)
# Encircling Mechanism
d = self.pop[idx].local_solution - self.pop[idx].solution
d_g = self.g_best.solution - self.pop[idx].solution
r1 = self.generator.random(self.problem.n_dims)
r2 = self.generator.random(self.problem.n_dims)
x_m = (1 - r1 - r2) * self.pop[idx].solution + r1 * d + r2 * d_g
x_m = self.correct_solution(x_m)
y_m = self.get_target(x_m)
# Discriminant Condition
y_list_compare = [y_0.fitness, y_n.fitness, y_m.fitness]
y_list = [y_0, y_n, y_m]
x_list = [x_0, x_n, x_m]
if self.problem.minmax == "min":
id_best = np.argmin(y_list_compare)
x_best = x_list[id_best]
y_best = y_list[id_best]
else:
id_best = np.argmax(y_list_compare)
x_best = x_list[id_best]
y_best = y_list[id_best]
if self.compare_target(y_best, self.pop[idx].target, self.problem.minmax):
self.pop[idx].solution = x_best
self.pop[idx].target = y_best
if self.compare_target(y_best, self.pop[idx].local_target, self.problem.minmax):
self.pop[idx].local_solution = x_best
self.pop[idx].local_target = y_best
self.pop[idx].g = (np.sum(self.pop[idx].weights * self.pop[idx].solution) - self.pop[idx].target.fitness) * self.pop[idx].solution
else:
if self.generator.random() < 0.3:
self.pop[idx].solution = x_best
self.pop[idx].target = y_best