# !/usr/bin/env python
# Created by "Thieu" at 14:51, 17/03/2020 ----------%
# Email: nguyenthieu2102@gmail.com %
# Github: https://github.com/thieu1995 %
# --------------------------------------------------%
import numpy as np
from copy import deepcopy
from mealpy.optimizer import Optimizer
[docs]class BaseSFO(Optimizer):
"""
The original version of: SailFish Optimizer (SFO)
Links:
1. https://doi.org/10.1016/j.engappai.2019.01.001
Hyper-parameters should fine tuned in approximate range to get faster convergen toward the global optimum:
+ pp (float): the rate between SailFish and Sardines (N_sf = N_s * pp) = 0.25, 0.2, 0.1
+ AP (int): A = 4, 6,... (coefficient for decreasing the value of Attack Power linearly from AP to 0)
+ epxilon (float): should be 0.0001, 0.001
Examples
~~~~~~~~
>>> import numpy as np
>>> from mealpy.swarm_based.SFO import BaseSFO
>>>
>>> def fitness_function(solution):
>>> return np.sum(solution**2)
>>>
>>> problem_dict1 = {
>>> "fit_func": fitness_function,
>>> "lb": [-10, -15, -4, -2, -8],
>>> "ub": [10, 15, 12, 8, 20],
>>> "minmax": "min",
>>> "verbose": True,
>>> }
>>>
>>> epoch = 1000
>>> pop_size = 50
>>> pp = 0.1
>>> AP = 4
>>> epxilon = 0.0001
>>> model = BaseSFO(problem_dict1, epoch, pop_size, pp, AP, epxilon)
>>> best_position, best_fitness = model.solve()
>>> print(f"Solution: {best_position}, Fitness: {best_fitness}")
References
~~~~~~~~~~
[1] Shadravan, S., Naji, H.R. and Bardsiri, V.K., 2019. The Sailfish Optimizer: A novel
nature-inspired metaheuristic algorithm for solving constrained engineering optimization
problems. Engineering Applications of Artificial Intelligence, 80, pp.20-34.
"""
def __init__(self, problem, epoch=10000, pop_size=100, pp=0.1, AP=4, epxilon=0.0001, **kwargs):
"""
Args:
problem (dict): The problem dictionary
epoch (int): maximum number of iterations, default = 10000
pop_size (int): number of population size, default = 100, SailFish pop size
pp (float): the rate between SailFish and Sardines (N_sf = N_s * pp) = 0.25, 0.2, 0.1
AP (int): A = 4, 6,... (coefficient for decreasing the value of Power Attack linearly from AP to 0)
epxilon (float): should be 0.0001, 0.001
"""
super().__init__(problem, kwargs)
self.nfe_per_epoch = 2 * pop_size
self.sort_flag = True
self.epoch = epoch
self.pop_size = pop_size
self.pp = pp
self.AP = AP
self.epxilon = epxilon
self.s_size = int(self.pop_size / self.pp)
[docs] def initialization(self):
self.pop = self.create_population(self.pop_size)
self.s_pop = self.create_population(self.s_size)
_, self.g_best = self.get_global_best_solution(self.pop) # pop = sailfish
_, self.s_gbest = self.get_global_best_solution(self.s_pop) # s_pop = sardines
[docs] def evolve(self, epoch):
"""
The main operations (equations) of algorithm. Inherit from Optimizer class
Args:
epoch (int): The current iteration
"""
## Calculate lamda_i using Eq.(7)
## Update the position of sailfish using Eq.(6)
nfe_epoch = 0
pop_new = []
PD = 1 - self.pop_size / (self.pop_size + self.s_size)
for i in range(0, self.pop_size):
lamda_i = 2 * np.random.uniform() * PD - PD
pos_new = self.s_gbest[self.ID_POS] - lamda_i * (np.random.uniform() *
(self.pop[i][self.ID_POS] + self.s_gbest[self.ID_POS]) / 2 - self.pop[i][self.ID_POS])
pos_new = self.amend_position(pos_new)
pop_new.append([pos_new, None])
self.pop = self.update_fitness_population(pop_new)
nfe_epoch += self.pop_size
## Calculate AttackPower using Eq.(10)
AP = self.AP * (1 - 2 * (epoch + 1) * self.epxilon)
if AP < 0.5:
alpha = int(self.s_size * np.abs(AP))
beta = int(self.problem.n_dims * np.abs(AP))
### Random np.random.choice number of sardines which will be updated their position
list1 = np.random.choice(range(0, self.s_size), alpha)
for i in range(0, self.s_size):
if i in list1:
#### Random np.random.choice number of dimensions in sardines updated, remove third loop by numpy vector computation
pos_new = deepcopy(self.s_pop[i][self.ID_POS])
list2 = np.random.choice(range(0, self.problem.n_dims), beta, replace=False)
pos_new[list2] = (np.random.uniform(0, 1, self.problem.n_dims) *
(self.pop[self.ID_POS] - self.s_pop[i][self.ID_POS] + AP))[list2]
pos_new = self.amend_position(pos_new)
self.s_pop[i] = [pos_new, None]
else:
### Update the position of all sardine using Eq.(9)
for i in range(0, self.s_size):
pos_new = np.random.uniform() * (self.g_best[self.ID_POS] - self.s_pop[i][self.ID_POS] + AP)
self.s_pop[i][self.ID_POS] = self.amend_position(pos_new)
## Recalculate the fitness of all sardine
self.s_pop = self.update_fitness_population(self.s_pop)
nfe_epoch += self.s_size
## Sort the population of sailfish and sardine (for reducing computational cost)
self.pop, g_best = self.get_global_best_solution(self.pop)
self.s_pop, s_gbest = self.get_global_best_solution(self.s_pop)
for i in range(0, self.pop_size):
for j in range(0, self.s_size):
### If there is a better position in sardine population.
if self.compare_agent(self.s_pop[j], self.pop[i]):
self.pop[i] = deepcopy(self.s_pop[j])
del self.s_pop[j]
break #### This simple keyword helped reducing ton of comparing operation.
#### Especially when sardine pop size >> sailfish pop size
temp = self.s_size - len(self.s_pop)
if temp == 1:
self.s_pop = self.s_pop + [self.create_solution()]
else:
self.s_pop = self.s_pop + self.create_population(self.s_size - len(self.s_pop))
_, self.s_gbest = self.get_global_best_solution(self.s_pop)
self.nfe_per_epoch = nfe_epoch
[docs]class ImprovedSFO(Optimizer):
"""
My improved version of: Sailfish Optimizer (I-SFO)
Notes
~~~~~
+ Reforms Energy equation
+ Removes parameters AP (A) and epsilon
+ Applies the idea of Opposition-based Learning technique
Hyper-parameters should fine tuned in approximate range to get faster convergen toward the global optimum:
+ pp (float): the rate between SailFish and Sardines (N_sf = N_s * pp) = 0.25, 0.2, 0.1
Examples
~~~~~~~~
>>> import numpy as np
>>> from mealpy.swarm_based.SFO import ImprovedSFO
>>>
>>> def fitness_function(solution):
>>> return np.sum(solution**2)
>>>
>>> problem_dict1 = {
>>> "fit_func": fitness_function,
>>> "lb": [-10, -15, -4, -2, -8],
>>> "ub": [10, 15, 12, 8, 20],
>>> "minmax": "min",
>>> "verbose": True,
>>> }
>>>
>>> epoch = 1000
>>> pop_size = 50
>>> pp = 0.1
>>> model = ImprovedSFO(problem_dict1, epoch, pop_size, pp)
>>> best_position, best_fitness = model.solve()
>>> print(f"Solution: {best_position}, Fitness: {best_fitness}")
References
~~~~~~~~~~
[1] Socha, K. and Dorigo, M., 2008. Ant colony optimization for continuous domains.
European journal of operational research, 185(3), pp.1155-1173.
"""
def __init__(self, problem, epoch=10000, pop_size=100, pp=0.1, **kwargs):
"""
Args:
problem (dict): The problem dictionary
epoch (int): maximum number of iterations, default = 10000
pop_size (int): number of population size, default = 100, SailFish pop size
pp (float): the rate between SailFish and Sardines (N_sf = N_s * pp) = 0.25, 0.2, 0.1
"""
super().__init__(problem, kwargs)
self.nfe_per_epoch = 2 * pop_size
self.sort_flag = True
self.epoch = epoch
self.pop_size = pop_size
self.pp = pp
self.s_size = int(self.pop_size / self.pp)
[docs] def initialization(self):
self.pop = self.create_population(self.pop_size)
self.s_pop = self.create_population(self.s_size)
_, self.g_best = self.get_global_best_solution(self.pop)
_, self.s_gbest = self.get_global_best_solution(self.s_pop)
[docs] def evolve(self, epoch):
"""
The main operations (equations) of algorithm. Inherit from Optimizer class
Args:
epoch (int): The current iteration
"""
## Calculate lamda_i using Eq.(7)
## Update the position of sailfish using Eq.(6)
nfe_epoch = 0
pop_new = []
for i in range(0, self.pop_size):
PD = 1 - len(self.pop) / (len(self.pop) + len(self.s_pop))
lamda_i = 2 * np.random.uniform() * PD - PD
pos_new = self.s_gbest[self.ID_POS] - lamda_i * (np.random.uniform() *
(self.g_best[self.ID_POS] + self.s_gbest[self.ID_POS]) / 2 - self.pop[i][self.ID_POS])
pos_new = self.amend_position(pos_new)
pop_new.append([pos_new, None])
self.pop = self.update_fitness_population(pop_new)
nfe_epoch += self.pop_size
## ## Calculate AttackPower using my Eq.thieu
#### This is our proposed, simple but effective, no need A and epxilon parameters
AP = 1 - epoch * 1.0 / self.epoch
if AP < 0.5:
for i in range(0, len(self.s_pop)):
temp = (self.g_best[self.ID_POS] + AP) / 2
pos_new = self.problem.lb + self.problem.ub - temp + np.random.uniform() * (temp - self.s_pop[i][self.ID_POS])
self.s_pop[i][self.ID_POS] = self.amend_position(pos_new)
else:
### Update the position of all sardine using Eq.(9)
for i in range(0, len(self.s_pop)):
pos_new = np.random.uniform() * (self.g_best[self.ID_POS] - self.s_pop[i][self.ID_POS] + AP)
self.s_pop[i][self.ID_POS] = self.amend_position(pos_new)
## Recalculate the fitness of all sardine
self.s_pop = self.update_fitness_population(self.s_pop)
nfe_epoch += len(self.s_pop)
## Sort the population of sailfish and sardine (for reducing computational cost)
self.pop = self.get_sorted_strim_population(self.pop, self.pop_size)
self.s_pop = self.get_sorted_strim_population(self.s_pop, len(self.s_pop))
for i in range(0, self.pop_size):
for j in range(0, len(self.s_pop)):
### If there is a better position in sardine population.
if self.compare_agent(self.s_pop[j], self.pop[i]):
self.pop[i] = deepcopy(self.s_pop[j])
del self.s_pop[j]
break #### This simple keyword helped reducing ton of comparing operation.
#### Especially when sardine pop size >> sailfish pop size
self.s_pop = self.s_pop + self.create_population(self.s_size - len(self.s_pop))
_, self.s_gbest = self.get_global_best_solution(self.s_pop)
self.nfe_per_epoch = nfe_epoch