#!/usr/bin/env python
# Created by "Thieu" at 22:07, 07/04/2020 ----------%
# Email: nguyenthieu2102@gmail.com %
# Github: https://github.com/thieu1995 %
# --------------------------------------------------%
import numpy as np
from mealpy.optimizer import Optimizer
[docs]class OriginalFA(Optimizer):
"""
The original version of: Fireworks Algorithm (FA)
Links:
1. https://doi.org/10.1007/978-3-642-13495-1_44
Hyper-parameters should fine-tune in approximate range to get faster convergence toward the global optimum:
+ max_sparks (int): parameter controlling the total number of sparks generated by the pop_size fireworks, default=100
+ p_a (float): percent (const parameter), default=0.04
+ p_b (float): percent (const parameter), default=0.8
+ max_ea (int): maximum explosion amplitude, default=40
+ m_sparks (int): number of sparks generated in each explosion generation, default=100
Examples
~~~~~~~~
>>> import numpy as np
>>> from mealpy import FloatVar, FA
>>>
>>> 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 = FA.OriginalFA(epoch=1000, pop_size=50, max_sparks = 50, p_a = 0.04, p_b = 0.8, max_ea = 40, m_sparks = 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] Tan, Y. and Zhu, Y., 2010, June. Fireworks algorithm for optimization. In International
conference in swarm intelligence (pp. 355-364). Springer, Berlin, Heidelberg.
"""
def __init__(self, epoch: int = 10000, pop_size: int = 100, max_sparks: int = 100, p_a: float = 0.04,
p_b: float = 0.8, max_ea: int = 40, m_sparks: int = 100, **kwargs: object) -> None:
"""
Args:
epoch (int): maximum number of iterations, default = 10000
pop_size (int): number of population size, default = 100
max_sparks (int): parameter controlling the total number of sparks generated by the pop_size fireworks, default=100
p_a (float): percent (const parameter), default=0.04
p_b (float): percent (const parameter), default=0.8
max_ea (int): maximum explosion amplitude, default=40
m_sparks (int): number of sparks generated in each explosion generation, default=100
"""
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.max_sparks = self.validator.check_int("max_sparks", max_sparks, [2, 10000])
self.p_a = self.validator.check_float("p_a", p_a, (0, 1.0))
self.p_b = self.validator.check_float("p_b", p_b, (0, 1.0))
self.max_ea = self.validator.check_int("max_ea", max_ea, [2, 100])
self.m_sparks = self.validator.check_int("m_sparks", m_sparks, [2, 10000])
self.set_parameters(["epoch", "pop_size", "max_sparks", "p_a", "p_b", "max_ea", "m_sparks"])
self.sort_flag = False
[docs] def evolve(self, epoch):
"""
The main operations (equations) of algorithm. Inherit from Optimizer class
Args:
epoch (int): The current iteration
"""
fit_list = np.array([agent.target.fitness for agent in self.pop])
fit_list = sorted(fit_list)
pop_new = []
for idx in range(0, self.pop_size):
si = self.max_sparks * (fit_list[-1] - self.pop[idx].target.fitness + self.EPSILON) / \
(self.pop_size * fit_list[-1] - np.sum(fit_list) + self.EPSILON)
Ai = self.max_ea * (self.pop[idx].target.fitness - fit_list[0] + self.EPSILON) / \
(np.sum(fit_list) - fit_list[0] + self.EPSILON)
if si < self.p_a * self.max_sparks:
si_ = int(round(self.p_a * self.max_sparks) + 1)
elif si > self.p_b * self.m_sparks:
si_ = int(round(self.p_b * self.max_sparks) + 1)
else:
si_ = int(round(si) + 1)
## Algorithm 1
pop_new = []
for j in range(0, si_):
pos_new = self.pop[idx].solution.copy()
list_idx = self.generator.choice(range(0, self.problem.n_dims), round(self.generator.uniform() * self.problem.n_dims), replace=False)
displacement = Ai * self.generator.uniform(-1, 1)
pos_new[list_idx] = pos_new[list_idx] + displacement
pos_new = np.where(np.logical_or(pos_new < self.problem.lb, pos_new > self.problem.ub),
self.problem.lb + np.abs(pos_new) % (self.problem.ub - self.problem.lb), pos_new)
pos_new = self.correct_solution(pos_new)
agent = self.generate_empty_agent(pos_new)
pop_new.append(agent)
if self.mode not in self.AVAILABLE_MODES:
pop_new[-1].target = self.get_target(pos_new)
pop_new = self.update_target_for_population(pop_new)
for _ in range(0, self.m_sparks):
idx = self.generator.integers(0, self.pop_size)
pos_new = self.pop[idx].solution.copy()
list_idx = self.generator.choice(range(0, self.problem.n_dims), round(self.generator.uniform() * self.problem.n_dims), replace=False)
pos_new[list_idx] = pos_new[list_idx] + self.generator.normal(0, 1, len(list_idx)) # Gaussian
condition = np.logical_or(pos_new < self.problem.lb, pos_new > self.problem.ub)
pos_true = self.problem.lb + np.abs(pos_new) % (self.problem.ub - self.problem.lb)
pos_new = np.where(condition, pos_true, pos_new)
pos_new = self.correct_solution(pos_new)
agent = self.generate_empty_agent(pos_new)
pop_new.append(agent)
if self.mode not in self.AVAILABLE_MODES:
pop_new[-1].target = self.get_target(pos_new)
pop_new = self.update_target_for_population(pop_new)
## Update the global best
self.pop = self.get_sorted_and_trimmed_population(pop_new + self.pop, self.pop_size, self.problem.minmax)