def _log_prob(self, counts):
counts = self._assert_valid_sample(counts)
log_unnormalized_prob = math_ops.reduce_sum(
counts * math_ops.log(self.p),
reduction_indices=[-1])
log_normalizer = -distribution_util.log_combinations(self.n, counts)
return log_unnormalized_prob - log_normalizer
def log_prob(self, counts, name="log_prob"):
"""`Log(P[counts])`, computed for every batch member.
For each batch of counts `[n_1,...,n_k]`, `P[counts]` is the probability
that after sampling `n` draws from this Multinomial distribution, the
number of draws falling in class `j` is `n_j`. Note that different
sequences of draws can result in the same counts, thus the probability
includes a combinatorial coefficient.
Args:
counts: Non-negative tensor with dtype `dtype` and whose shape can
be broadcast with `self.p` and `self.n`. For fixed leading dimensions,
the last dimension represents counts for the corresponding Multinomial
distribution in `self.p`. `counts` is only legal if it sums up to `n`
and its components are equal to integer values.
name: Name to give this Op, defaults to "log_prob".
Returns:
Log probabilities for each record, shape `[N1,...,Nm]`.
"""
n = self._n
p = self._p
with ops.name_scope(self.name):
with ops.op_scope([n, p, counts], name):
counts = self._check_counts(counts)
prob_prob = math_ops.reduce_sum(counts * math_ops.log(self._p),
reduction_indices=[-1])
log_prob = prob_prob + distribution_util.log_combinations(
n, counts)
return log_prob
def _log_prob(self, counts):
counts = self._maybe_assert_valid_sample(counts)
ordered_prob = (
special_math_ops.lbeta(self.concentration + counts)
- special_math_ops.lbeta(self.concentration))
return ordered_prob + distribution_util.log_combinations(
self.total_count, counts)
def _log_prob(self, counts):
counts = self._assert_valid_counts(counts)
ordered_prob = (special_math_ops.lbeta(self.alpha + counts) -
special_math_ops.lbeta(self.alpha))
log_prob = ordered_prob + distribution_util.log_combinations(
self.n, counts)
return log_prob
def log_prob(self, counts, name="log_prob"):
"""`Log(P[counts])`, computed for every batch member.
For each batch of counts `[n_1,...,n_k]`, `P[counts]` is the probability
that after sampling `n` draws from this Dirichlet Multinomial
distribution, the number of draws falling in class `j` is `n_j`. Note that
different sequences of draws can result in the same counts, thus the
probability includes a combinatorial coefficient.
Args:
counts: Non-negative tensor with dtype `dtype` and whose shape can be
broadcast with `self.alpha`. For fixed leading dimensions, the last
dimension represents counts for the corresponding Dirichlet Multinomial
distribution in `self.alpha`. `counts` is only legal if it sums up to
`n` and its components are equal to integer values.
name: Name to give this Op, defaults to "log_prob".
Returns:
Log probabilities for each record, shape `[N1,...,Nn]`.
"""
n = self._n
alpha = self._alpha
with ops.name_scope(self.name):
with ops.name_scope(name, values=[n, alpha, counts]):
counts = self._check_counts(counts)
ordered_prob = (special_math_ops.lbeta(alpha + counts) -
special_math_ops.lbeta(alpha))
log_prob = ordered_prob + distribution_util.log_combinations(
n, counts)
return log_prob
def _log_prob(self, counts):
counts = self._assert_valid_counts(counts)
ordered_prob = (special_math_ops.lbeta(self.alpha + counts) -
special_math_ops.lbeta(self.alpha))
log_prob = ordered_prob + distribution_util.log_combinations(
self.n, counts)
return log_prob
def log_prob(self, counts, name="log_prob"):
"""`Log(P[counts])`, computed for every batch member.
For each batch of counts `[n_1,...,n_k]`, `P[counts]` is the probability
that after sampling `n` draws from this Multinomial distribution, the
number of draws falling in class `j` is `n_j`. Note that different
sequences of draws can result in the same counts, thus the probability
includes a combinatorial coefficient.
Args:
counts: Non-negative tensor with dtype `dtype` and whose shape can
be broadcast with `self.p` and `self.n`. For fixed leading dimensions,
the last dimension represents counts for the corresponding Multinomial
distribution in `self.p`. `counts` is only legal if it sums up to `n`
and its components are equal to integer values.
name: Name to give this Op, defaults to "log_prob".
Returns:
Log probabilities for each record, shape `[N1,...,Nm]`.
"""
n = self._n
p = self._p
with ops.name_scope(self.name):
with ops.op_scope([n, p, counts], name):
counts = self._check_counts(counts)
prob_prob = math_ops.reduce_sum(counts * math_ops.log(self._p),
reduction_indices=[-1])
log_prob = prob_prob + distribution_util.log_combinations(
n, counts)
return log_prob
def log_prob(self, counts, name="log_prob"):
"""`Log(P[counts])`, computed for every batch member.
For each batch of counts `[n_1,...,n_k]`, `P[counts]` is the probability
that after sampling `n` draws from this Dirichlet Multinomial
distribution, the number of draws falling in class `j` is `n_j`. Note that
different sequences of draws can result in the same counts, thus the
probability includes a combinatorial coefficient.
Args:
counts: Non-negative tensor with dtype `dtype` and whose shape can be
broadcast with `self.alpha`. For fixed leading dimensions, the last
dimension represents counts for the corresponding Dirichlet Multinomial
distribution in `self.alpha`. `counts` is only legal if it sums up to
`n` and its components are equal to integer values.
name: Name to give this Op, defaults to "log_prob".
Returns:
Log probabilities for each record, shape `[N1,...,Nn]`.
"""
n = self._n
alpha = self._alpha
with ops.name_scope(self.name):
with ops.name_scope(name, values=[n, alpha, counts]):
counts = self._check_counts(counts)
ordered_prob = (special_math_ops.lbeta(alpha + counts) -
special_math_ops.lbeta(alpha))
log_prob = ordered_prob + distribution_util.log_combinations(
n, counts)
return log_prob
def testLogCombinationsShape(self):
# Shape [2, 2]
n = [[2, 5], [12, 15]]
with self.test_session():
n = np.array(n, dtype=np.float32)
# Shape [2, 2, 4]
counts = [[[1., 1, 0, 0], [2., 2, 1, 0]], [[4., 4, 1, 3], [10, 1, 1, 4]]]
log_binom = distribution_util.log_combinations(n, counts)
self.assertEqual([2, 2], log_binom.get_shape())
def testLogCombinationsShape(self):
# Shape [2, 2]
n = [[2, 5], [12, 15]]
with self.test_session():
n = np.array(n, dtype=np.float32)
# Shape [2, 2, 4]
counts = [[[1., 1, 0, 0], [2., 2, 1, 0]], [[4., 4, 1, 3], [10, 1, 1, 4]]]
log_binom = distribution_util.log_combinations(n, counts)
self.assertEqual([2, 2], log_binom.get_shape())
def testLogCombinationsBinomial(self):
n = [2, 5, 12, 15]
k = [1, 2, 4, 11]
log_combs = np.log(special.binom(n, k))
with self.test_session():
n = np.array(n, dtype=np.float32)
counts = [[1., 1], [2., 3], [4., 8], [11, 4]]
log_binom = distribution_util.log_combinations(n, counts)
self.assertEqual([4], log_binom.get_shape())
self.assertAllClose(log_combs, log_binom.eval())
def testLogCombinationsBinomial(self):
n = [2, 5, 12, 15]
k = [1, 2, 4, 11]
log_combs = np.log(special.binom(n, k))
with self.test_session():
n = np.array(n, dtype=np.float32)
counts = [[1., 1], [2., 3], [4., 8], [11, 4]]
log_binom = distribution_util.log_combinations(n, counts)
self.assertEqual([4], log_binom.get_shape())
self.assertAllClose(log_combs, log_binom.eval())
def _log_prob(self, counts):
counts = self._maybe_assert_valid_sample(counts)
ordered_prob = (special_math_ops.lbeta(self.concentration + counts) -
special_math_ops.lbeta(self.concentration))
return ordered_prob + distribution_util.log_combinations(
self.total_count, counts)
def _log_normalization(self, counts):
counts = self._maybe_assert_valid_sample(counts)
return -distribution_util.log_combinations(self.total_count, counts)
def _log_normalization(self, counts): counts = self._maybe_assert_valid_sample(counts) return -distribution_util.log_combinations(self.total_count, counts)
