plt.title('initial')
plt.savefig(file_path + '0.png')
num_episodes = 500
k = 1
for i in xrange(num_episodes):
r_t, s_t = domain.reset()
agent.reset()
count = 0
cumulative_reward = 0
while s_t != None:
# apply an action from the agent
# the domain will return a 'None' state when terminating
r_t, s_t = domain.step(agent.step(r_t, s_t))
count += 1
cumulative_reward += r_t
# final update step for the agent
agent.step(r_t, s_t)
if i % 2 == 0:
if render_value_fn:
plt.gca().clear()
plt.contourf(*getValueFn(valuefn))
plt.title('episode ' + str(i))
plt.savefig(file_path + str(k) + '.png')
k +=1
# print cumulative reward it took to reach the goal
