def __init__(self, question, test_dict):
super(QLearningTest, self).__init__(question, test_dict)
self.discount = float(test_dict['discount'])
self.grid = grid_world.GridWorld(parse_grid(test_dict['grid']))
if 'noise' in test_dict:
self.grid.set_noise(float(test_dict['noise']))
if 'living_reward' in test_dict:
self.grid.set_living_reward(float(test_dict['living_reward']))
self.grid = grid_world.GridWorld(parse_grid(test_dict['grid']))
self.env = grid_world.GridWorldEnvironment(self.grid)
self.epsilon = float(test_dict['epsilon'])
self.learning_rate = float(test_dict['learning_rate'])
self.opts = {
'action_fn': self.env.get_possible_actions,
'epsilon': self.epsilon,
'gamma': self.discount,
'alpha': self.learning_rate
}
num_experiences = int(test_dict['num_experiences'])
max_pre_experiences = 10
self.nums_experiences_for_display = range(
min(num_experiences, max_pre_experiences))
self.test_out_file = test_dict['test_out_file']
if max_pre_experiences < num_experiences:
self.nums_experiences_for_display.append(num_experiences)
def __init__(self, question, test_dict):
super(EpsilonGreedyTest, self).__init__(question, test_dict)
self.discount = float(test_dict['discount'])
self.grid = grid_world.GridWorld(parse_grid(test_dict['grid']))
if 'noise' in test_dict:
self.grid.set_noise(float(test_dict['noise']))
if 'living_reward' in test_dict:
self.grid.set_living_reward(float(test_dict['living_reward']))
self.grid = grid_world.GridWorld(parse_grid(test_dict['grid']))
self.env = grid_world.GridWorldEnvironment(self.grid)
self.epsilon = float(test_dict['epsilon'])
self.learning_rate = float(test_dict['learning_rate'])
self.num_experiences = int(test_dict['num_experiences'])
self.num_iterations = int(test_dict['iterations'])
self.opts = {
'action_fn': self.env.get_possible_actions,
'epsilon': self.epsilon,
'gamma': self.discount,
'alpha': self.learning_rate
}
return "get_" + all_cap_re.sub(r'\1_\2', s1).lower()
if __name__ == '__main__':
opts = parse_options()
###########################
# GET THE GRIDWORLD
###########################
import grid_world
mdp_function = getattr(grid_world, get_function(opts.grid))
mdp = mdp_function()
mdp.set_living_reward(opts.living_reward)
mdp.set_noise(opts.noise)
env = grid_world.GridWorldEnvironment(mdp)
###########################
# GET THE DISPLAY ADAPTER
###########################
import text_grid_world_display
display = text_grid_world_display.TextGridWorldDisplay(mdp)
if not opts.text_display:
import graphics_grid_world_display
display = graphics_grid_world_display.GraphicsGridWorldDisplay(mdp, opts.grid_size, opts.speed)
try:
display.start()
except KeyboardInterrupt:
sys.exit(0)
###########################
# GET THE AGENT