def createExperimentInstance():
gymRawEnv = gym.make('MountainCarContinuous-v0')
cartPositionGroup = Digitizer.buildBins(-1.2, 0.6, 16)
cartVelocityGroup = Digitizer.buildBins(-0.07, 0.07, 4)
actionDedigitizer = Digitizer.build(-1.0, 1.0, 5, True)
# print("Cart position bins:", cartPositionGroup)
# print("Cart velocity bins:", cartVelocityGroup)
# print("Cart force bins:", actionDedigitizer.bins, actionDedigitizer.possibleValues())
observationDigitizer = ArrayDigitizer(
[cartPositionGroup, cartVelocityGroup])
transformation = EnvTransformation(observationDigitizer, actionDedigitizer)
task = GymTask.createTask(gymRawEnv)
env = task.env
env.setTransformation(transformation)
# env.setCumulativeRewardMode()
# create agent with controller and learner - use SARSA(), Q() or QLambda() here
## alpha -- learning rate (preference of new information)
## gamma -- discount factor (importance of future reward)
# create value table and initialize with ones
table = ActionValueTable(observationDigitizer.states,
actionDedigitizer.states)
table.initialize(0.0)
# table.initialize( np.random.rand( table.paramdim ) )
agent = createAgent(table)
experiment = Experiment(task, agent)
experiment = ProcessExperiment(experiment, doSingleExperiment)
return experiment
def test_values_noEdges_5(self):
digitizer = Digitizer.build(-1.0, 1.0, 5, False)
npt.assert_array_almost_equal(digitizer.values, [-0.6, -0.3, 0.0, 0.3, 0.6], 3)
def test_build_edge_5(self):
digitizer = Digitizer.build(0.0, 10.0, 5, True)
self.assertEqual(digitizer.states, 5)
npt.assert_array_almost_equal(digitizer.bins, [0., 3.333, 6.667, 10.], 3)
npt.assert_array_almost_equal(digitizer.values, [0., 2.5, 5., 7.5, 10.], 3)
def test_values_edges_5(self):
digitizer = Digitizer.build(-1.0, 1.0, 5, True)
npt.assert_array_almost_equal(digitizer.values, [-1.0, -0.5, 0.0, 0.5, 1.0], 3)
def test_build_noEdge_5(self):
digitizer = Digitizer.build(0.0, 10.0, 5, False)
self.assertEqual(digitizer.states, 5)
npt.assert_array_almost_equal(digitizer.bins, [2., 4., 6., 8.], 3)
npt.assert_array_almost_equal(digitizer.values, [2., 3.5, 5., 6.5, 8.], 3)