Merge remote-tracking branch 'origin/aprimoramentos'

Example.py

@@ -3,25 +3,31 @@
 Example of main program to be used with PSO
 @author: Daniel
 """
+from matplotlib import use
+use('Agg')
+
 from PSO import PSO # Importing the PSO class
 from Modelo_benchmark import Modelo # Importing the function to be minimized
+from time import time
 
-sup = [32,32] # maximum value for the parameters
-inf = [-32,-32] # minimum value for the parameters
+t1 = time()
 
+sup = [100,100] # maximum value for the parameters
+inf = [-100,-100] # minimum value for the parameters
 
 args_model=['ackley'] # In this example, the Modelo class needs another argument: the name of the function to be minimized
-
 # PSO Algorithm executed with decreasing inercia weight from 0.9 to 0.4; acceleration factors constant and equal to default values of 2; 30 particles and 200 iterations.
-Otimizacao = PSO(sup,inf,{'busca':'Otimo','algoritmo':'PSO','inercia':'TVIW-linear','aceleracao':'Constante','restricao':True},\
-Num_particulas=30,itmax=100,w=[0.9,0.4],args_model=args_model)
+Otimizacao = PSO(sup,inf,{'busca':'Otimo','algoritmo':'PSO','inercia':'TVIW-Adaptive-vel','aceleracao':'TVAC','restricao':True,'parada':['itmax']},
+Num_particulas=30,itmax=500,w=[1.0,0.1],args_model=args_model,NP=2)
+#Otimizacao = PSO(sup,inf,{'busca':'Otimo','algoritmo':'HPSO','gbest':'Particula','parada':'itmax','inercia':'Constante'},\
+# Num_particulas=30,itmax=100,w=[0.9,0.4],args_model=args_model)
 Otimizacao.Busca(Modelo,printit=True) # Do the search
-Otimizacao.Result_txt() # Printing results in the same folder as txt files
-Otimizacao.Graficos(Nome_param=[r'$\theta_1$',r'$\theta_2$'],Unid_param=[None,None],FO2a2=True)  # Creating performance graphs in the same folder, including the objective function in 3d (FO2a2=True)
-
+Otimizacao.Relatorios(resumos_txt=True) # Printing results in the same folder as txt files
+#Otimizacao.Graficos(Nome_param=[r'$\theta_{}$'.format(i) for i in xrange(len(sup))],FO2a2=True) # Creating performance graphs in the same folder, including the objective function in 3d (FO2a2=True)
 # To create movie frames uncomment the line below
-# Otimizacao.Movie(Nome_param=[r'$\theta_1$',r'$\theta_2$'],tipos=['projecao','funcao']) # Creating movie frames
+#Otimizacao.Movie(Nome_param=[r'$\theta_1$',r'$\theta_2$'],Unid_param=['admin','admin']) # Creating movie frames
 
 # Printing the results:
 print 'Optimum point', Otimizacao.gbest # optimum point
 print 'Best Fitness' , Otimizacao.best_fitness # objective function in the optimum point
+print 'Time', time()-t1

Modelo_benchmark.py

@@ -7,46 +7,49 @@
 import sys
 from numpy import exp, size, sin, cos, sqrt, pi, sum, matrix
 
+
 class Modelo(Thread):
     result = 0
-    def __init__(self,param,args_model):
+
+    def __init__(self, param, args_model):
         Thread.__init__(self)
         self.args = param
         self.args_model = args_model
-    
+
     def run(self):
-
-	#Funções unimodais
-
-	if self.args_model[0] == 'exponential':
-	    # Função Exponencial
-	    aux = -exp(-0.5*sum([arg**2 for arg in self.args]))
-
-	if self.args_model[0] == 'sphere':
-	    # Função sphere (f2)
-	    aux = sum([arg**2 for arg in self.args])
-
-	if self.args_model[0] == 'rosenbrook':
-	    # Função de Rosenbrock
-	    aux = 0
-	    for i in xrange(0,size(self.args)-1):
-	        aux = (100*(self.args[i+1]-self.args[i]**2)**2 + (self.args[i]-1)**2) + aux
-
-	#Funções multimodais
-
-	if self.args_model[0] == 'shaffer_f6':
-	    # Shaffer's f6
-	    num = sin(sqrt(self.args[0]**2 + self.args[1]**2))**2 - 0.5
-	    den = (1.0 + 0.001*(self.args[0]**2 + self.args[1]**2))**2
-	    aux = 0.5 + num/den
-
-	if self.args_model[0] == 'rastrigin':
-	    # Rastrigin
-	    aux = sum([arg**2 - 10*cos(2*pi*arg) + 10 for arg in self.args])
-
-	if self.args_model[0] == 'ackley':
-	    # Ackley
-	    D = float(size(self.args))
-	    aux = -20*exp(-0.2*sqrt((1/D)*sum([arg**2 for arg in self.args]))) -exp((1/D)*sum(cos([2*pi*arg for arg in self.args]))) + 20 + exp(1)
-
-	self.result = aux
+
+        # Funções unimodais
+
+        if self.args_model[0] == 'exponential':
+            # Função Exponencial
+            aux = -exp(-0.5 * sum([arg ** 2 for arg in self.args]))
+
+        if self.args_model[0] == 'sphere':
+            # Função sphere (f2)
+            aux = sum([arg ** 2 for arg in self.args])
+
+        if self.args_model[0] == 'rosenbrook':
+            # Função de Rosenbrock
+            aux = 0
+            for i in xrange(0, size(self.args) - 1):
+                aux = (100 * (self.args[i + 1] - self.args[i] ** 2) ** 2 + (self.args[i] - 1) ** 2) + aux
+
+        # Funções multimodais
+
+        if self.args_model[0] == 'shaffer_f6':
+            # Shaffer's f6
+            num = sin(sqrt(self.args[0] ** 2 + self.args[1] ** 2)) ** 2 - 0.5
+            den = (1.0 + 0.001 * (self.args[0] ** 2 + self.args[1] ** 2)) ** 2
+            aux = 0.5 + num / den
+
+        if self.args_model[0] == 'rastrigin':
+            # Rastrigin
+            aux = sum([arg ** 2 - 10 * cos(2 * pi * arg) + 10 for arg in self.args])
+
+        if self.args_model[0] == 'ackley':
+            # Ackley
+            D = float(size(self.args))
+            aux = -20 * exp(-0.2 * sqrt((1 / D) * sum([arg ** 2 for arg in self.args]))) - exp(
+                (1 / D) * sum(cos([2 * pi * arg for arg in self.args]))) + 20 + exp(1)
+
+        self.result = aux