def test_from_dict_with_hmm(self):
for (starting_order, num_states), model in sorted(self.models.items()):
if starting_order >= 5:
continue
state_priors = numpy.random.random(model.low_order_states)
state_priors /= state_priors.sum()
trans_probs = numpy.random.random(
(model.low_order_states, model.low_order_states))
trans_probs = (trans_probs.T / trans_probs.sum(1)).T
my_hmm = FirstOrderHMM(state_priors=ArrayFactor(state_priors),
trans_probs=MatrixFactor(trans_probs),
emission_probs=[None] *
model.low_order_states)
dtmp = {
"starting_order": starting_order,
"num_states": num_states,
"hmm": my_hmm,
}
found = ModelReducer._from_dict(dtmp)
expected = ModelReducer(starting_order, num_states)
expected.hmm = my_hmm
for k in ("starting_order", "high_order_states"):
yield check_equal, getattr(found, k), getattr(expected, k)
for k in ("trans_probs", "state_priors"):
yield check_array_equal, getattr(found.hmm,
k).data, getattr(my_hmm,
k).data
def test_from_dict_no_hmm(self):
for (starting_order, num_states), model in sorted(self.models.items()):
dtmp = {
"starting_order": starting_order,
"num_states": num_states,
}
found = ModelReducer._from_dict(dtmp)
expected = ModelReducer(starting_order, num_states)
for k in ("starting_order", "high_order_states"):
yield check_equal, getattr(found, k), getattr(expected, k)
yield check_none, found._hmm
def test_first_order_is_identical(self):
expected = {(0, ): 0, (1, ): 1, (2, ): 2, (3, ): 3}
model = ModelReducer(1, 4)
with warnings.catch_warnings():
warnings.simplefilter("ignore", UserWarning)
assert_dict_equal(model.high_states_to_low, expected)
assert_dict_equal(model.low_states_to_high, self.revdict(expected))
def test_transcode_sequence(self):
testseq = ["A", "B", "C", "D", "E"]
expected = numpy.arange(5)
dtmp = {K: V for K, V in zip(testseq, expected)}
found = ModelReducer.transcode_sequence(testseq, dtmp)
yield check_array_equal, found, expected
def test_remap_from_first_order(self):
# test parameter remapping by asserting that log probabilities of observation sequences
# from remapped high-order mdodels match those of low order models
models = {
2: get_dirty_casino(),
4: get_fourstate_poisson(),
}
for num_states, native_model in sorted(models.items()):
seqs = [native_model.generate(200)[1] for _ in range(5)]
for starting_order in 2, 3, 4, 5:
mod = ModelReducer(starting_order, num_states)
trans_model = mod.remap_from_first_order(native_model)
for my_obs in seqs:
native_logprob = native_model.logprob(my_obs)
trans_logprob = native_model.logprob(my_obs)
yield check_almost_equal, native_logprob, trans_logprob
def test_transcode_sequences(self):
testseqs = [["A", "B", "C", "D", "E"], ["D", "B", "A", "A"]]
expected = [numpy.arange(5), numpy.array([3, 1, 0, 0])]
dtmp = {K: V for K, V in zip(testseqs[0], expected[0])}
found = ModelReducer.transcode_sequences(testseqs, dtmp)
yield check_equal, len(found), len(expected)
for my_found, my_expected in zip(found, expected):
yield check_array_equal, my_found, my_expected
def test_get_emission_mapping(self):
cases = {
(2, 2): numpy.tile(range(2), 3),
(2, 3): numpy.tile(range(3), 4),
(2, 4): numpy.tile(range(4), 5),
(3, 2): numpy.tile(range(2), 7),
(3, 3): numpy.tile(range(3), 13),
(3, 4): numpy.tile(range(4), 21),
(4, 2): numpy.tile(range(2), 15),
(4, 3): numpy.tile(range(3), 40),
(4, 4): numpy.tile(range(4), 85),
}
for (order, states), expected in sorted(cases.items()):
found = ModelReducer(order, states).get_emission_mapping()
yield check_array_equal, expected, found
def setUpClass(cls):
cls.models = {}
cls.max_order = 6
cls.max_states = 7
for num_states in range(2, cls.max_states):
for starting_order in range(1, cls.max_order):
with warnings.catch_warnings():
warnings.simplefilter("ignore")
model = ModelReducer(starting_order, num_states)
cls.models[(starting_order, num_states)] = model
cls.sequences = [
[0, 2, 0, 5, 2, 2, 4, 2, 4, 1],
]
# expected tuples for each seq in cls.sequences
# when moving in model order K
cls.expected_tuples = {
2: [
[(-1, 0), (0, 2), (2, 0), (0, 5), (5, 2), (2, 2), (2, 4),
(4, 2), (2, 4), (4, 1)],
],
3: [
[
(-2, -1, 0),
(-1, 0, 2),
(0, 2, 0),
(2, 0, 5),
(0, 5, 2),
(5, 2, 2),
(2, 2, 4),
(2, 4, 2),
(4, 2, 4),
(2, 4, 1),
],
],
4: [
[
(-3, -2, -1, 0),
(-2, -1, 0, 2),
(-1, 0, 2, 0),
(0, 2, 0, 5),
(2, 0, 5, 2),
(0, 5, 2, 2),
(5, 2, 2, 4),
(2, 2, 4, 2),
(2, 4, 2, 4),
(4, 2, 4, 1),
],
],
5: [
[
(-4, -3, -2, -1, 0),
(-3, -2, -1, 0, 2),
(-2, -1, 0, 2, 0),
(-1, 0, 2, 0, 5),
(0, 2, 0, 5, 2),
(2, 0, 5, 2, 2),
(0, 5, 2, 2, 4),
(5, 2, 2, 4, 2),
(2, 2, 4, 2, 4),
(2, 4, 2, 4, 1),
],
],
}
# keys are (num_states, model_order)
cls.expected_high_states_to_low = {
(4, 2): {
# created by adding new start state
(-1, 0): 0,
(-1, 1): 1,
(-1, 2): 2,
(-1, 3): 3,
# represented
(0, 0): 4,
(0, 1): 5,
(0, 2): 6,
(0, 3): 7,
(1, 0): 8,
(1, 1): 9,
(1, 2): 10,
(1, 3): 11,
(2, 0): 12,
(2, 1): 13,
(2, 2): 14,
(2, 3): 15,
(3, 0): 16,
(3, 1): 17,
(3, 2): 18,
(3, 3): 19,
},
(4, 3): {
# created from new start states
(-2, -1, 0): 0,
(-2, -1, 1): 1,
(-2, -1, 2): 2,
(-2, -1, 3): 3,
(-1, 0, 0): 4,
(-1, 0, 1): 5,
(-1, 0, 2): 6,
(-1, 0, 3): 7,
(-1, 1, 0): 8,
(-1, 1, 1): 9,
(-1, 1, 2): 10,
(-1, 1, 3): 11,
(-1, 2, 0): 12,
(-1, 2, 1): 13,
(-1, 2, 2): 14,
(-1, 2, 3): 15,
(-1, 3, 0): 16,
(-1, 3, 1): 17,
(-1, 3, 2): 18,
(-1, 3, 3): 19,
# actual states
(0, 0, 0): 20,
(0, 0, 1): 21,
(0, 0, 2): 22,
(0, 0, 3): 23,
(0, 1, 0): 24,
(0, 1, 1): 25,
(0, 1, 2): 26,
(0, 1, 3): 27,
(0, 2, 0): 28,
(0, 2, 1): 29,
(0, 2, 2): 30,
(0, 2, 3): 31,
(0, 3, 0): 32,
(0, 3, 1): 33,
(0, 3, 2): 34,
(0, 3, 3): 35,
(1, 0, 0): 36,
(1, 0, 1): 37,
(1, 0, 2): 38,
(1, 0, 3): 39,
(1, 1, 0): 40,
(1, 1, 1): 41,
(1, 1, 2): 42,
(1, 1, 3): 43,
(1, 2, 0): 44,
(1, 2, 1): 45,
(1, 2, 2): 46,
(1, 2, 3): 47,
(1, 3, 0): 48,
(1, 3, 1): 49,
(1, 3, 2): 50,
(1, 3, 3): 51,
(2, 0, 0): 52,
(2, 0, 1): 53,
(2, 0, 2): 54,
(2, 0, 3): 55,
(2, 1, 0): 56,
(2, 1, 1): 57,
(2, 1, 2): 58,
(2, 1, 3): 59,
(2, 2, 0): 60,
(2, 2, 1): 61,
(2, 2, 2): 62,
(2, 2, 3): 63,
(2, 3, 0): 64,
(2, 3, 1): 65,
(2, 3, 2): 66,
(2, 3, 3): 67,
(3, 0, 0): 68,
(3, 0, 1): 69,
(3, 0, 2): 70,
(3, 0, 3): 71,
(3, 1, 0): 72,
(3, 1, 1): 73,
(3, 1, 2): 74,
(3, 1, 3): 75,
(3, 2, 0): 76,
(3, 2, 1): 77,
(3, 2, 2): 78,
(3, 2, 3): 79,
(3, 3, 0): 80,
(3, 3, 1): 81,
(3, 3, 2): 82,
(3, 3, 3): 83
},
}
def test_revive_from_json(self):
for k, v in self.models.items():
enc = v.to_json()
dec = ModelReducer.from_json(enc)
yield check_equal, v, dec