# !/usr/bin/env python
# Created by "Thieu" at 17:48, 18/03/2020 ----------%
# Email: nguyenthieu2102@gmail.com %
# Github: https://github.com/thieu1995 %
# --------------------------------------------------%
import numpy as np
from mealpy.optimizer import Optimizer
[docs]class BaseHS(Optimizer):
"""
My changed version of: Harmony Search (HS)
Links:
1. https://doi.org/10.1177/003754970107600201
Notes
~~~~~
- Used the global best in the harmony memories
- Removed third for loop
Hyper-parameters should fine tuned in approximate range to get faster convergen toward the global optimum:
+ c_r (float): [0.1, 0.5], Harmony Memory Consideration Rate), default = 0.15
+ pa_r (float): [0.3, 0.8], Pitch Adjustment Rate, default=0.5
Examples
~~~~~~~~
>>> import numpy as np
>>> from mealpy.music_based.HS import BaseHS
>>>
>>> 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
>>> c_r = 0.95
>>> pa_r = 0.05
>>> model = BaseHS(problem_dict1, epoch, pop_size, c_r, pa_r)
>>> best_position, best_fitness = model.solve()
>>> print(f"Solution: {best_position}, Fitness: {best_fitness}")
"""
def __init__(self, problem, epoch=10000, pop_size=100, c_r=0.95, pa_r=0.05, **kwargs):
"""
Args:
problem (dict): The problem dictionary
epoch (int): maximum number of iterations, default = 10000
pop_size (int): number of population size, default = 100
c_r (float): Harmony Memory Consideration Rate, default = 0.15
pa_r (float): Pitch Adjustment Rate, default=0.5
"""
super().__init__(problem, kwargs)
self.nfe_per_epoch = pop_size
self.sort_flag = False
self.epoch = epoch
self.pop_size = pop_size
self.c_r = c_r
self.pa_r = pa_r
self.fw = 0.0001 * (self.problem.ub - self.problem.lb) # Fret Width (Bandwidth)
self.fw_damp = 0.9995 # Fret Width Damp Ratio
self.dyn_fw = self.fw
[docs] def evolve(self, epoch):
"""
The main operations (equations) of algorithm. Inherit from Optimizer class
Args:
epoch (int): The current iteration
"""
pop_new = []
for idx in range(0, self.pop_size):
# Create New Harmony Position
pos_new = np.random.uniform(self.problem.lb, self.problem.ub)
delta = self.dyn_fw * np.random.normal(self.problem.lb, self.problem.ub)
# Use Harmony Memory
pos_new = np.where(np.random.uniform(0, 1, self.problem.n_dims) < self.c_r, self.g_best[self.ID_POS], pos_new)
# Pitch Adjustment
x_new = pos_new + delta
pos_new = np.where(np.random.uniform(0, 1, self.problem.n_dims) < self.pa_r, x_new, pos_new)
pos_new = self.amend_position(pos_new) # Check the bound
pop_new.append([pos_new, None])
pop_new = self.update_fitness_population(pop_new)
# Update Damp Fret Width
self.dyn_fw = self.dyn_fw * self.fw_damp
# Merge Harmony Memory and New Harmonies, Then sort them, Then truncate extra harmonies
self.pop = self.get_sorted_strim_population(self.pop + pop_new, self.pop_size)
[docs]class OriginalHS(BaseHS):
"""
The original version of: Harmony Search (HS)
Links:
1. https://doi.org/10.1177/003754970107600201
Hyper-parameters should fine tuned in approximate range to get faster convergen toward the global optimum:
+ c_r (float): [0.1, 0.5], Harmony Memory Consideration Rate), default = 0.15
+ pa_r (float): [0.3, 0.8], Pitch Adjustment Rate, default=0.5
Examples
~~~~~~~~
>>> import numpy as np
>>> from mealpy.music_based.HS import OriginalHS
>>>
>>> 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
>>> c_r = 0.95
>>> pa_r = 0.05
>>> model = OriginalHS(problem_dict1, epoch, pop_size, c_r, pa_r)
>>> best_position, best_fitness = model.solve()
>>> print(f"Solution: {best_position}, Fitness: {best_fitness}")
References
~~~~~~~~~~
[1] Geem, Z.W., Kim, J.H. and Loganathan, G.V., 2001. A new heuristic
optimization algorithm: harmony search. simulation, 76(2), pp.60-68.
"""
def __init__(self, problem, epoch=10000, pop_size=100, c_r=0.95, pa_r=0.05, **kwargs):
"""
Args:
problem (dict): The problem dictionary
epoch (int): maximum number of iterations, default = 10000
pop_size (int): number of population size, default = 100
c_r (float): Harmony Memory Consideration Rate), default = 0.15
pa_r (float): Pitch Adjustment Rate, default=0.5
"""
super().__init__(problem, epoch, pop_size, c_r, pa_r, **kwargs)
self.nfe_per_epoch = pop_size
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
"""
pop_new = []
for idx in range(0, self.pop_size):
pos_new = np.random.uniform(self.problem.lb, self.problem.ub)
for j in range(self.problem.n_dims):
# Use Harmony Memory
if np.random.uniform() <= self.c_r:
random_index = np.random.randint(0, self.pop_size)
pos_new[j] = self.pop[random_index][self.ID_POS][j]
# Pitch Adjustment
if np.random.uniform() <= self.pa_r:
delta = self.dyn_fw * np.random.normal(self.problem.lb, self.problem.ub) # Gaussian(Normal)
pos_new[j] = pos_new[j] + delta[j]
pos_new = self.amend_position(pos_new)
pop_new.append([pos_new, None])
pop_new = self.update_fitness_population(pop_new)
# Update Damp Fret Width
self.dyn_fw = self.dyn_fw * self.fw_damp
# Merge Harmony Memory and New Harmonies, Then sort them, Then truncate extra harmonies
self.pop = self.get_sorted_strim_population(self.pop + pop_new, self.pop_size)