def bubbleSort(data):
data_length = len(data)
for i in range(data_length - 1):
for j in range(data_length - i - 1):
if data[j] > data[j + 1]:
data[j], data[j + 1] = data[j + 1], data[j]
draw(j + 1, data)
def selectionSort(data):
data_length = len(data)
for i in range(data_length - 1):
least = i
for j in range(i + 1, data_length):
if data[j] < data[least]:
least = j
if i != least:
data[i], data[least] = data[least], data[i]
draw(i + 1, data)
def insertionSort(data):
data_length = len(data)
for i in range(1, data_length):
j = i - 1
key = data[i]
while data[j] > key and j >= 0:
data[j + 1] = data[j]
j = j - 1
data[j + 1] = key
draw(j + 1, data)
def heapify(data, index, size):
largest = index
left = 2 * index + 1
right = 2 * index + 2
if left < size and data[left] > data[largest]:
largest = left
if right < size and data[right] > data[largest]:
largest = right
if largest != index:
data[largest], data[index] = data[index], data[largest]
heapify(data, largest, size)
draw(largest, data)
def testQLearning(gamma, epsilon, surprise):
global state_sequence
print '** Testing Q-Learning algorithm **'
print 'Gamma: ', gamma
print 'Epsilon: ', epsilon
print 'Suprise: ', surprise
qLearning(gamma, epsilon, surprise)
l = len(state_sequence)
print '\tState sequence ', state_sequence[l - 20:]
animate.draw(state_sequence[l - 20:])
def testQLearning(gamma, epsilon, surprise): global state_sequence print '** Testing Q-Learning algorithm **' print 'Gamma: ', gamma print 'Epsilon: ', epsilon print 'Suprise: ', surprise qLearning(gamma, epsilon, surprise) l = len(state_sequence) print '\tState sequence ', state_sequence[l-20:] animate.draw(state_sequence[l-20:])
def main():
# Get the variables used to do reinforcement learning
gamma = 0.7
T = 10000
epsilon = 0.2
s = 12
n = 0.1
e = Environment(s)
Q = calculate_route(T, n, gamma, e, s, epsilon)
print("calculation done")
draw(e.get_loop(Q, s))
def main():
# Get the variables used to do reinforcement learning
gamma = 0.7
T = 10000
epsilon = 0.2
s = 12
n = 0.1
e = Environment(s)
Q = calculate_route(T, n, gamma, e, s, epsilon)
print("calculation done")
draw(e.get_loop(Q, s))
def testPolicyIteration(rew, gamma, iterations): print '** Testing policy iteration algorithm **' print 'Gamma: ', gamma print 'Iterations: ', iterations policyIteration(rew, gamma, iterations) print '\tOptimal policy: ', policy optimal_state_map = [ trans[a][policy[a]] for a in range(len(policy)) ] print '\tOptimal state map', optimal_state_map current_state = int(random.random()*len(policy)) path = [current_state] for i in range(20): current_state = optimal_state_map[current_state] path += [current_state] print '\t(Hopefully) correct walking path', path animate.draw(path)
def testPolicyIteration(rew, gamma, iterations):
print '** Testing policy iteration algorithm **'
print 'Gamma: ', gamma
print 'Iterations: ', iterations
policyIteration(rew, gamma, iterations)
print '\tOptimal policy: ', policy
optimal_state_map = [trans[a][policy[a]] for a in range(len(policy))]
print '\tOptimal state map', optimal_state_map
current_state = int(random.random() * len(policy))
path = [current_state]
for i in range(20):
current_state = optimal_state_map[current_state]
path += [current_state]
print '\t(Hopefully) correct walking path', path
animate.draw(path)
a = policy[s]
print str(rew[s][a]) + "= " + str(s) + "|" + str(a)
print str(rew[s][a]) + " + " + str(gamma) + " * " + str(value[trans[s][a]])
value[s] = rew[s][a] + gamma * value[trans[s][a]]
def move(state, steps):
ret = [state]
for s in range(steps):
state = trans[state][policy[state]]
ret.append(state)
return ret
print rew
print "=== before "
print policy
print value
pol_iter(100)
print "=== after "
print policy
print value
result = move(0, 20)
a.draw(result)
pylab.imread('step5.png'),
pylab.imread('step6.png'),
pylab.imread('step7.png'),
pylab.imread('step8.png'),
pylab.imread('step9.png'),
pylab.imread('step10.png'),
pylab.imread('step11.png'),
pylab.imread('step12.png'),
pylab.imread('step13.png'),
pylab.imread('step14.png'),
pylab.imread('step15.png'),
pylab.imread('step16.png'))
#comic = numpy.concatenate([images[i] for i in result], axis=1)
#pylab.imshow(comic)
#pylab.show()
env = Environment()
epsilon = 0.01
eta = 0.1
gamma = 0.9
q = qlearn(env, 200000, epsilon, eta, gamma)
print(q)
moves= qMove(0, 20, q)
print(moves)
a.draw(moves)
comic = numpy.concatenate([images[i] for i in moves], axis=1)
pylab.imshow(comic)
pylab.show()
pylab.imread('step9.png'),
pylab.imread('step10.png'),
pylab.imread('step11.png'),
pylab.imread('step12.png'),
pylab.imread('step13.png'),
pylab.imread('step14.png'),
pylab.imread('step15.png'),
pylab.imread('step16.png'))
#visualization of the robot walk
#comic = numpy.concatenate([images[i] for i in sequence], axis=1)
#pylab.imshow(comic)
#pylab.show()
animate.draw(sequence)
class Environment :
'''
Representation of the environment for the Q-learning algorithm
'''
def __init__(self, state=0):
self.state = state
self.trans = ((1, 3 ,4 ,12),
(0, 2, 5, 13),
(3, 1, 6, 14),
(2, 0, 7, 15),
(5, 7, 0, 8),
(4, 6, 1, 9),
