def __init__(self,
model,
k=2,
discount_rate=0.9,
u0=1,
std0=0.2,
sampling_interval=20):
self.model = model
self.discount_rate = discount_rate
self.keepr = Keeper()
# priority queue for ML
self.ML_queue = UniquePriorityQueue()
# comparison constant
self.delta = 0.001
# number of back-up per action
self.k = k
# default mean and std
self.u0 = u0
self.std0 = std0
# draw initial hypothesis
self.hypothesis = Hypothesis.draw_init_hypothesis(
model, self.u0, self.std0)
# maximum-likelihood V
self.ML_V = {}
# interval to draw samples
self.sampling_interval = sampling_interval
def test_1(self):
queue = UniquePriorityQueue()
queue.push(2, 3)
queue.push(0, 5)
queue.push(8, 1)
(p, v) = queue.pop()
self.assertEqual(5, v)
(p, v) = queue.pop()
self.assertEqual(3, v)
(p, v) = queue.pop()
self.assertEqual(1, v)
def test_2(self):
queue = UniquePriorityQueue()
queue.push_or_update(2, 3)
queue.push_or_update(10, 5)
queue.push_or_update(8, 1)
queue.push_or_update(0, 5)
queue.push_or_update(4, 6)
(p, v) = queue.pop()
self.assertEqual(5, v)
(p, v) = queue.pop()
self.assertEqual(3, v)
(p, v) = queue.pop()
self.assertEqual(6, v)
(p, v) = queue.pop()
self.assertEqual(1, v)
queue.push_or_update(10, 5)
(p, v) = queue.pop()
self.assertEqual(5, v)
def __init__(self,
model,
k=2,
epsilon=1,
degrading_constant=0.99,
discount_rate=0.9):
self.model = model
# value model
self.V = {}
# book-keeping keeper
self.keepr = Keeper()
# parameters for the algorithm
self.k = k
self.epsilon = epsilon
self.degrading_constant = degrading_constant
self.discount_rate = discount_rate
# priority queue
self.queue = UniquePriorityQueue()
self.delta = 0.001
