def updateBat(bat, bmin, bmax, best, fmin, fmax, A, G, dim, size, alpha = 0.8, gamma = 0.9):
"""bat.frequency = fmin + (fmax - fmin) * random.uniform(0,1)
bat.velocity = list(map(operator.add, bat.velocity, (_ * bat.frequency for _ in map(operator.sub, bat, best))))
solution = creator.Bat(list(map(operator.add, bat , bat.velocity)))"""
rand = random.uniform(0,1)#numpy.random.random_sample()
#print("best random: ",rand,"the rate copared to it: ",bat.rate)
if rand > bat.rate: #random walk using equation 5 from the paper on the global best solution.
eps = random.uniform(-1,1)
eps_A = eps * A
solution = numpy.add(best,eps_A)#( _ + eps_A for _ in best)
solution = creator.Bat(list(solution))
"""print("-around the best-")
print("eps: ",eps)
print("A: ",A)
print("solution: ",solution)"""
else:
bat.frequency = [[fmin + (fmax - fmin) * random.uniform(0,1) for _ in range(dim)] for _ in range(size)] #equation 2 from the paper.
dis = numpy.multiply(numpy.subtract(bat, best), bat.frequency)
bat.velocity = numpy.add(bat.velocity, dis) #eauqtion 3 from the paper.
solution = creator.Bat(numpy.add(bat , bat.velocity)) #equation 4 from the paper
"""print("-random walk-")
print("frequency: ",bat.frequency)
print("velocity: ",bat.velocity)
print("solution: ",solution)"""
for _ in solution: #making sure the bat doesn't go too far and stays in the objective function domain.
for __ in range(len(_)):
if _[__] > bmax:
_[__] = bmax
if _[__] < bmin:
_[__] = bmin
#print("solution after boudning: ",solution)
solution.fitness = toolbox.evaluate(data = histogram , bat = solution)#calculation the fitness of the Bat.
rand = random.uniform(0,1)#numpy.random.random_sample()
#print("best random: ",rand,"the loudness copared to it: ",bat.loudness,rand < bat.loudness)
if bat.fitness > solution.fitness and rand < bat.loudness: #asserting the solution only if the solution is better and the bat iss too loud.
#print("-asserting the solution- rand:",rand,"loudness:",bat.loudness)
bat[:] = solution
bat.fitness = solution.fitness
bat.loudness = alpha * bat.loudness
bat.rate = bat.init_rate * (1 - math.exp(-alpha * (G+1)))
#print("new loudness",bat.loudness)
#print("new rate",bat.rate)
if bat.best.fitness > bat.fitness: #updating the personal best.
#print("pbest updated")
bat.best = creator.Bat(bat)
bat.best.fitness = bat.fitness
if best.fitness > solution.fitness: #updating the global best.
#print("best updated")
best[:] = creator.Bat(solution)
best.fitness = solution.fitness
def generate(bmin, bmax, dim, size):
bat = creator.Bat([random.uniform(bmin,bmax), random.uniform(bmin,numpy.amax(histogram))] for _ in range(size))
bat.init_rate = random.uniform(0,0.95)
bat.rate = bat.init_rate
bat.loudness = 1
bat.velocity = [[0,0] for _ in range(size)]
bat.fitness = toolbox.evaluate(data = histogram , bat = bat)
bat.best = creator.Bat(bat)
bat.best.fitness = bat.fitness
return bat
def updateBat(bat,
best,
fmin,
fmax,
G,
A,
m,
histogram,
size,
dim,
alpha=0.9,
gamma=0.9):
bat.frequency = fmin + (fmax - fmin) * numpy.random.uniform(0, 1)
dis = numpy.multiply(numpy.subtract(bat, best), bat.frequency)
# eauqtion 3 from the paper.
bat.velocity = list(numpy.add(bat.velocity, dis))
# equation 4 from the paper
solution = creator.Bat(list(numpy.add(bat, bat.velocity)))
rand = numpy.random.random_sample()
# random walk using equation 5 from the paper on the global best solution.
if rand > bat.rate:
solution = numpy.add(best,
[[random.uniform(-1, 1) * A for _ in range(dim)]
for _ in range(size)])
#solution = numpy.add(best , numpy.multiply( A ,[random.gauss(0,1), random.gauss(0,1)]))
solution = creator.Bat(list(solution))
for i in range(len(solution)):
for _, x in enumerate(solution[i]):
if solution[i][0] > 255:
print("top changed", solution[i][0])
solution[i][0] = 255
print("top changed")
if x < 0:
solution[i][_] = 0
print("x")
# calculation the fitness of the Bat.
solution.fitness = toolbox.evaluate(
solution, m,
histogram=histogram) # calculation the fitness of the Bat.
rand = numpy.random.random_sample()
# asserting the solution only if the solution is better and the bat is too loud.
if bat.fitness >= solution.fitness and rand < bat.loudness:
bat[:] = solution
bat.fitness = solution.fitness
bat.loudness = alpha * bat.loudness
bat.rate = bat.init_rate * (1 - math.exp(-gamma * (G + 1)))
if best.fitness > solution.fitness: # updating the global best.
best[:] = creator.Bat(solution)
best.fitness = solution.fitness
def main():
Population = toolbox.population(n=20)
"""print(Population)
print(" ")"""
GEN = 1500
best = None
A = 0
for bat in Population: #finding the initial global best
if not best or best.fitness > bat.fitness:
best = creator.Bat(bat)
best.fitness = bat.fitness
A += bat.loudness
"""print("the bat: ",bat , "the fitness: ", bat.fitness,"the loudness: ",bat.loudness,"the rate: ",bat.rate,"initrate",bat.init_rate, )
print(" ")"""
mean_A = A / len(Population) #calculating the average loudness to use in equation 5 of the paper.
for G in range(GEN): #computing the BA GEN times
A=0
"""print("the best:", best, "fitness: ", best.fitness)
print(" ")
print("average loudness:",mean_A)"""
for bat in Population: #updating the Bat postion one by one and opdating the Global best as well.
toolbox.update(bat, best = best, A = mean_A, G = G)
A += bat.loudness
"""print("the bat: ",bat , "the fitness: ", bat.fitness,"the loudness: ",bat.loudness,"the rate: ",bat.rate,"initrate",bat.init_rate, )
print(" ")"""
mean_A = A / len(Population) #calculating the new average loudness
"""for bat in Population:
if best.fitness > bat.fitness:
best = creator.Bat(bat)
best.fitness = bat.fitness"""
print(best, best.fitness)
def updateBat(bat,
best,
fmin,
fmax,
G,
A,
m,
histogram,
size,
dim,
alpha=0.9,
gamma=0.9):
bat.frequency = fmin + (fmax - fmin) * numpy.random.uniform(0, 1)
dis = numpy.multiply(numpy.subtract(bat, best), bat.frequency)
bat.velocity = list(numpy.add(bat.velocity,
dis)) #eauqtion 3 from the paper.
solution = creator.Bat(list(numpy.add(
bat, bat.velocity))) #equation 4 from the paper
rand = numpy.random.random_sample()
if rand > bat.rate: #random walk using equation 5 from the paper on the global best solution.
solution = numpy.add(best,
[[random.uniform(-1, 1) * A for _ in range(dim)]
for _ in range(size)])
#solution = numpy.add(best , numpy.multiply( A ,[random.gauss(0,1), random.gauss(0,1)]))
solution = creator.Bat(list(solution))
""" for _ in range(len(solution)): #making sure the bat doesn't go too far and stays in the objective function domain.
if solution[_] > bmax:
solution[_] = bmax
if solution[_] < bmin:
solution[_] = bmin
"""
solution.fitness = toolbox.evaluate(
solution, m, histogram=histogram) #calculation the fitness of the Bat.
rand = numpy.random.random_sample()
if bat.fitness >= solution.fitness and rand < bat.loudness: #asserting the solution only if the solution is better and the bat is too loud.
bat[:] = solution
bat.fitness = solution.fitness
bat.loudness = alpha * bat.loudness
bat.rate = bat.init_rate * (1 - math.exp(-gamma * (G + 1)))
if best.fitness > solution.fitness: #updating the global best.
best[:] = creator.Bat(solution)
best.fitness = solution.fitness
def generate(bmin, bmax, m, histogram, size, init_R, init_A):
bat = creator.Bat([[
random.uniform(bmin, bmax),
random.uniform(bmin, numpy.amax(histogram))
] for _ in range(size)])
bat.init_rate = random.uniform(0, init_R)
bat.rate = bat.init_rate
bat.loudness = random.uniform(init_A, 2)
bat.velocity = [[0, 0] for _ in range(size)]
bat.fitness = toolbox.evaluate(bat, m, histogram=histogram)
return bat
