class SmartHome(Home):
def __init__(self, w, h):
super().__init__(w,h)
self.scenario = Scenario()
#self.presence, self.bulbs = self.scenario.diagonal(self.width, self.height)
#self.presence, self.bulbs = self.scenario.stripes(self.width, self.height)
self.presence, self.bulbs = self.scenario.corners(self.width, self.height)
self.init_deap()
self.init_figures()
self.pop = self.toolbox.population(n=1000)
def init_figures(self):
self.fig = plt.figure(figsize=(1, 3))
self.fig.add_subplot(131)
plt.imshow(self.presence, cmap='gray', interpolation='nearest', vmin=0, vmax=1)
self.fig.add_subplot(132)
plt.imshow(self.bulbs, cmap='gray', interpolation='nearest', vmin=-1, vmax=0)
self.fig.add_subplot(133)
self.im = plt.imshow(self.bulbs, cmap='gray', interpolation='bilinear', animated=True, vmin=0, vmax=1)
def init_deap(self):
creator.create("FitnessMax", base.Fitness, weights=(1.0,))
creator.create("Individual", list, fitness=creator.FitnessMax)
self.toolbox = base.Toolbox()
self.toolbox.register(
"random_num",
random.choice,
[0,1])
self.DIM = self.width * self.height
self.toolbox.register(
"individual",
tools.initRepeat,
creator.Individual,
self.toolbox.random_num,
n=self.DIM)
print (self.toolbox.individual())
self.toolbox.register("population",
tools.initRepeat,
list,
self.toolbox.individual)
self.toolbox.register("mate", tools.cxTwoPoint)
self.toolbox.register("select", tools.selBest)
self.toolbox.register("evaluate", self.evaluateInd4)
self.toolbox.register("mutate", self.myMutation)
def fitness(self, individual, data):
print (data)
print (individual)
match = 0
#for selected, bulbs in zip(individual, data):
#if selected:
for x in range(self.width):
for y in range(self.height):
if self.presence[x,y]>0 and individual[x,y]>0:
match += 1
match_percentage = match / self.width * self.height
return match_percentage
def evaluateInd(self, individual):
match = 0
#for selected, bulbs in zip(individual, data):
#if selected:
for y in range(self.height): # top left is (X=0,Y=0)
for x in range(self.width):
if individual[y*self.width+x]>0 and self.presence[x,y]>0:
match += 1
match_percentage = match / (self.width * self.height)
return (float(match_percentage),)
def evaluateInd2(self, individual):
match = 0
#for selected, bulbs in zip(individual, data):
#if selected:
for y in range(self.height): # top left is (X=0,Y=0)
for x in range(self.width):
if individual[y*self.width+x] == self.presence[x,y]:
if self.bulbs[x,y] > -1:
match += 1
else:
match -= 1 #penalty
match_percentage = match / (self.width * self.height)
return (float(match_percentage),)
def evaluateInd3(self, individual):
score = 0
#for selected, bulbs in zip(individual, data):
#if selected:
for y in range(self.height): # top left is (X=0,Y=0)
for x in range(self.width):
if individual[y*self.width+x]>0:
#if self.presence_in_radius(1, x, y):
presence_score = self.presence_in_radius(1, x, y)
if self.bulbs[x,y] > -1 and presence_score > 0:
score += self.presence_in_radius(1, x, y)
else:
score -= 1 #penalty for using a broken bulb
return (float(score),)
#match_percentage = match / (self.width * self.height)
#return (float(match_percentage),)
def presence_in_radius (self, radius, x, y):
block = ((x-1, y-1), (x, y-1), (x+1,y-1), (x+1, y), (x+1, y+1), (x, y+1), (x-1, y+1), (x-1, y)) # starts from left top
presence_count = 0
for point in block:
if point[0] in range(self.width):
if point[1] in range(self.height):
if self.presence[point[0], point[1]] > 0: #someone present
presence_count += 1
return presence_count
def evaluateInd4(self, individual):
score = 0
#for selected, bulbs in zip(individual, data):
#if selected:
for y in range(self.height): # top left is (X=0,Y=0)
for x in range(self.width):
if individual[y*self.width+x]>0:
#if self.presence_in_radius(1, x, y):
presence_score = self.presence_in_radius2(1, x, y, individual)
if self.bulbs[x,y] > -1 and presence_score > 0:
score += presence_score
else:
score -= 1 #penalty for using a broken bulb
return (float(score),)
def evaluateInd_table(self, individual):
#bulb-state, luminosity, presence : reward sign
logic_table = {
'000':1,
'001':-1,
'010':-1,
'011':1,
'-100':1,
'-101':1,
'-110':-1,
'-111':-1
}
score = 0
#for selected, bulbs in zip(individual, data):
#if selected:
for y in range(self.height): # top left is (X=0,Y=0)
for x in range(self.width):
presence_score = self.presence_in_radius2(1, x, y, individual)
key = ''
key += str(int(self.bulbs[x,y]))
key += str(int(individual[y*self.width+x])) #luninosity
key += str(int(1 if presence_score>0 else 0)) # presence
reward_sign = logic_table[key] #if logic_table[key] else 1
score = score + (reward_sign * 1)
return (float(score),)
def presence_in_radius2 (self, radius, x, y, individual):
block = ((x-1, y-1), (x, y-1), (x+1,y-1), (x+1, y), (x+1, y+1), (x, y+1), (x-1, y+1), (x-1, y)) # starts from left top
presence_count = 0
if self.presence[x,y] > 0:
presence_count += 50 #award for presence under the bulb
#print ('+50')
for point in block:
if point[0] in range(self.width):
if point[1] in range(self.height):
if self.presence[point[0], point[1]] > 0 and individual[point[1]*self.width+point[0]] < 1:#someone present in radius and that area is dark
presence_count += 1
return presence_count
def myMutation(self, individual):
luminosity = self.decode(individual)
for i in range(self.width):
x = random.randint(0, self.width-1)
y = random.randint(0, self.height-1)
if(self.bulbs[x,y] > -1):
luminosity[x,y] = random.choice([0,1])
#luminosity = self.encode (luminosity)
self.update_individual(individual, luminosity)
#for i in range(self.width):
# individual[random.randint(0,len(individual)-1)] = random.choice([0,1])
return (individual,)
def update_individual (self, individual, data):
for y in range(self.height):
for x in range(self.width):
individual[y*self.width+x] = data[x,y]
def encode(self, luminosity):
return luminosity.flatten()
def decode(self, individual):
bulbs = np.zeros((self.width,self.height))
for y in range(self.height): # top left is (X=0,Y=0)
for x in range(self.width):
bulbs[x,y] = individual[y*self.width+x]
return bulbs #np.reshape(individual, (-1, self.width))
def updatefig(self, *args):
algorithms.eaMuPlusLambda (
self.pop, self.toolbox,
400, 100, #parents, children
0.8, 0.2, #probabilities
1) #iterations
top = sorted(self.pop, key=lambda x:x.fitness.values[0])[-1]
fit = top.fitness.values[0]
print ('fitness-: {}'.format(fit))
self.im.set_data(self.decode(top))
#im.set_cmap("gray")
#im.update()
return self.im,
def run(self):
ani = animation.FuncAnimation(self.fig, self.updatefig, interval=50, blit=True)
plt.show()
