def __init__(self, func_id, pop_size=100, dim=10, max_iters=100):
self.dim = dim
self.pop_size = pop_size
self.xMin = -100 # Search space limits
self.xMax = 100 #
self.maxIters = max_iters
self.func_id = func_id
self.fitnessEvals = 0
# Original parameters
self.c1 = 2.0
self.c2 = 2.0
self.w = 1.0
self.wdamp = 1.0
# Yarpiz default
# self.c1 = 1.4962
# self.c2 = 1.4962
# self.w = 0.7298
# self.wdamp = 1.0
self.problem = pg.problem(
pg.cec2014(prob_id=self.func_id, dim=self.dim))
self.psoProblem = {
'CostFunction': self.get_fitness,
'nVar': self.dim,
'xMin': self.
xMin, # Alternatively you can use a "numpy array" with nVar elements, instead of scalar
'xMax': self.
xMax, # Alternatively you can use a "numpy array" with nVar elements, instead of scalar
}
def get_solution(func_id=1, dim=10, input_data_filepath=None):
'''
func_list: Function id, between 1 and 31.
dim : Problem dimensionality
Returns a solution corresponding the global optima of the function given
by func_id.
'''
import pygmo as pg
from glob import glob
# solution = dict()
if func_id < 23:
prob = pg.problem(pg.cec2014(prob_id=func_id, dim=dim))
filepath = dirs.input if input_data_filepath is None else input_data_filepath
shift_data = np.loadtxt(filepath +
"/shift_data_{}.txt".format(func_id))
# solution[func_id] = prob.fitness(shift_data[:dim])[0]
solution = prob.fitness(shift_data[:dim])[0]
if func_id >= 23:
raise NotImplementedError(
"f_{:2d} not yet implemented".format(func_id))
return None
return solution
def __init__(self, dim=2, func_id=1, pop_size=20):
self.dim = dim # Problem dimensionality
self.xMin = -100 # Search space limits
self.xMax = 100 #
self.pop_size = pop_size # Population size
self.func_id = func_id
self.fitnessEvals = 0 # Initial fitness evaluations
# Algorithm parameters
self.minSigma = 0.0001
self.tau1 = 4 * 1 / (np.sqrt(2 * self.pop_size))
self.tau2 = 4 * 1 / (np.sqrt(2 * np.sqrt(self.pop_size)))
self.mutateProb = 1.0
self.numTourneyPlays = 10
# Fitness Function definition
if self.func_id < 23:
self.problem = pg.problem(
pg.cec2014(prob_id=self.func_id, dim=self.dim))
else:
raise ValueError("f_{:2d} not yet implemented".format(
self.func_id))
return -1
# Initialize population DataFrame and compute initial Fitness
self.init_states()
def __init__(self, func_id, pop_size=100, dim=10, max_iters=100):
self.dim = dim
self.pop_size = pop_size
self.xMin = -100 # Search space limits
self.xMax = 100 #
self.vMin = self.xMin / 2 # Velocity limits
self.vMax = self.xMax / 2 #
self.func_id = func_id
self.fitnessEvals = 0
# # Debug
# self.c1 = 1.0
# self.c2 = 1.0
# self.w = 1.0
# self.wdamp = 1.0
# GOPSO default
self.c1 = 1.49618
self.c2 = 1.49618
self.w = 0.72984
self.wdamp = 1.0
self.problem = pg.problem(
pg.cec2014(prob_id=self.func_id, dim=self.dim))
self.init_states()
def __init__(self, dim=2, func_id=1, pop_size=30):
# Initialize parameters
self.dim = dim
self.pop_size = pop_size
self.xMin = -100 # Search space limits
self.xMax = 100 #
self.func_id = func_id
self.fitnessEvals = 0
# Parameters
self.param_F = 0.9 # Mutation parameter F
self.cross_rate = 0.9 # Crossover probability
# Fitness Function definition
self.problem = pg.problem(
pg.cec2014(prob_id=self.func_id, dim=self.dim))
# Initialize population DataFrame and compute initial Fitness
self.init_states()
def __init__(self, dim=2, func_id=1, pop_size=30):
# Initialize superclass EvolutionaryAlgorithm
super().__init__(dim=dim, pop_size=pop_size)
self.xMin = -100 # Search space limits
self.xMax = 100 #
self.func_id = func_id
# Parameters
self.param_F = [0.8, 0.9] # Mutation parameter F
# CURRENTLY MUST BE A LIST OF CHOICES FOR F
self.cross_rate = 0.9 # Crossover probability
# Fitness Function definition
if self.func_id < 23:
self.problem = pg.problem(
pg.cec2014(prob_id=self.func_id, dim=self.dim))
else:
raise NotImplementedError("f_{:2d} not yet implemented".format(
self.func_id))
return -1
# Initialize population DataFrame and compute initial Fitness
self.init_states()
""" Benchmark to test the cec2014 objective-functions
"""
import numpy as np
import pandas as pd
import pygmo as pg
import dirs
func_list = [1, 2, 6, 7, 9, 14] # Assignment function list
dim = 10
# Initializing a test vector with zero elements
x = np.zeros(dim)
for func_id in func_list:
if func_id < 23:
prob = pg.problem(pg.cec2014(prob_id=func_id, dim=dim))
shift_data = np.loadtxt(dirs.input +
"shift_data_{}.txt".format(func_id))
print("f_{:2d}(0) = {:.6f}".format(func_id, prob.fitness(x)[0]))
print("f_{:2d}(x*) = {:.6f}\n".format(
func_id,
prob.fitness(shift_data[:dim])[0]))
else:
print("f_{:2d} not yet implemented".format(func_id))
import pygmo as pg
import numpy as np
prob = pg.problem(pg.cec2014(prob_id=5, dim=10))
algo = pg.algorithm(pg.cmaes(gen=100))
archi = pg.archipelago(8, algo=algo, prob=prob, pop_size=20)
archi.evolve(1000)
archi.wait()
res = [isl.get_population().champion_f for isl in archi]
res = np.array(res)
print(f"Problem {5}: {res.mean()}")
def __init__( self, **kw ): import pygmo as pg dim = retrieve_kw( kw, 'dim', 2 ) prob = retrieve_kw( kw, 'prob', 1) self._cec2014_core = pg.problem(pg.cec2014( prob, dim ))
import numpy as np
import pandas as pd
import pygmo as pg
import dirs
# funcList = [1, 2, 6, 7, 9, 14] # Assignment function list
funcList = [23]
funcList = range(1, 31)
dim = 10
x = np.zeros(dim)
print("\n")
for i in funcList:
if i < 23:
prob = pg.problem(pg.cec2014(prob_id=i, dim=dim))
shift_data = np.loadtxt(dirs.input + "shift_data_{}.txt".format(i))
print("f_{:2d}(0) = {:.6f}".format(i, prob.fitness(x)[0]))
print("f_{:2d}(x*) = {:.6f}\n".format(
i,
prob.fitness(shift_data[:dim])[0]))
if i >= 23:
print("f_{:2d} not yet implemented".format(i))
