def search(self):
start = time.process_time()
count = 1
while True:
neighbor_states = self.action(self.current_state)
if len(neighbor_states) < 1:
print('Restart len 0')
self.current_state = self.env.random_start_state()
neighbor_states = self.action(self.current_state)
highest_state = get_neighbor_highest(neighbor_states)
highest_state.h = cal_heuristic(highest_state)
self.current_state.h = cal_heuristic(self.current_state)
if highest_state.h > self.current_state.h or self.goal(
self.current_state):
if self.current_state.h == 0:
self.current_state.show()
cal_heuristic(self.current_state)
if self.goal(self.current_state):
duration = time.process_time() - start
print(duration)
return self.current_state
else:
print('Restart stuck')
self.current_state = self.env.random_start_state()
else:
self.current_state = highest_state
self.current_state.h = cal_heuristic(self.current_state)
print('---------------------')
print('Time %s' % count)
print('H = %s' % self.current_state.h)
count += 1
def get_neighbor_highest(states):
state = states.pop()
min = cal_heuristic(state)
for other in states:
if cal_heuristic(other) < min:
min = cal_heuristic(other)
state = other
return state
def reproduce(father, mother):
child_h = min([father.h, mother.h])
child = State()
for i in range(5):
child = State()
c = random.randint(0, size - 1)
for column in range(c):
for row in range(size):
if father.map[row][column] == 1:
child.map[row][column] = 1
break
for column in range(c, size):
for row in range(size):
if mother.map[row][column] == 1:
child.map[row][column] = 1
break
child.h = cal_heuristic(child)
if child.h < child_h:
break
return child
def random_selection(population):
fitnessList = []
for state in population:
fitnessList.append(nCk(size, 2) - cal_heuristic(state))
probaList = []
total = sum(fitnessList)
first = 0
second = 0
for item in fitnessList:
probaList.append(item / total)
for i in range(len(probaList)):
if i > 0:
probaList[i] += probaList[i - 1]
random_value = random.uniform(0, 1)
for i in range(len(probaList)):
if random_value < probaList[i]:
first = i
break
random_value = random.uniform(0, 1)
for i in range(len(probaList)):
if random_value < probaList[i]:
second = i
break
return population[first], population[second]
def search(self):
count = 1
print(nCk(size, 2))
start = time.process_time()
while not self.is_best_individual():
fitness = ''
new_states = []
avg = 0
print('Time %s' % count)
for i in range(self.state_size):
father, mother = random_selection(self.states)
child = reproduce(father, mother)
if random.uniform(0, 1) < 0.3:
child = mutate(child)
new_states.append(child)
child.h = cal_heuristic(child)
avg += child.h
fitness += str(child.h) + ' '
avg = avg / self.state_size
list_childs = []
for item in new_states:
if item.h <= avg:
list_childs.append(item)
print(fitness)
self.states = list_childs
count += 1
duration = time.process_time() - start
print(duration)
for state in self.states:
if cal_heuristic(state) == 0:
return state
return None
def is_best_individual(self):
for state in self.states:
if cal_heuristic(state) == 0:
return True
return False
def __init__(self, env):
self.current_state = env.current_state
self.current_state.h = cal_heuristic(self.current_state)
self.goal = env.goal
self.action = env.action
self.env = env