def testPmfAndCdfNonIntegerCounts(self):
p = [[0.1, 0.2, 0.7]]
n = [[5.]]
with self.test_session():
# No errors with integer n.
binom = binomial.Binomial(total_count=n,
probs=p,
validate_args=True)
binom.prob([2., 3, 2]).eval()
binom.prob([3., 1, 2]).eval()
binom.cdf([2., 3, 2]).eval()
binom.cdf([3., 1, 2]).eval()
# Both equality and integer checking fail.
with self.assertRaisesOpError(
"cannot contain fractional components."):
binom.prob([1.0, 2.5, 1.5]).eval()
with self.assertRaisesOpError(
"cannot contain fractional components."):
binom.cdf([1.0, 2.5, 1.5]).eval()
binom = binomial.Binomial(total_count=n,
probs=p,
validate_args=False)
binom.prob([1., 2., 3.]).eval()
binom.cdf([1., 2., 3.]).eval()
# Non-integer arguments work.
binom.prob([1.0, 2.5, 1.5]).eval()
binom.cdf([1.0, 2.5, 1.5]).eval()
def testPmfAndCdfNonIntegerCounts(self):
p = [[0.1, 0.2, 0.7]]
n = [[5.]]
with self.cached_session():
# No errors with integer n.
binom = binomial.Binomial(total_count=n,
probs=p,
validate_args=True)
binom.prob([2., 3, 2]).eval()
binom.prob([3., 1, 2]).eval()
binom.cdf([2., 3, 2]).eval()
binom.cdf([3., 1, 2]).eval()
placeholder = array_ops.placeholder(dtypes.float32)
# Both equality and integer checking fail.
with self.assertRaisesOpError(
"cannot contain fractional components."):
binom.prob(placeholder).eval(
feed_dict={placeholder: [1.0, 2.5, 1.5]})
with self.assertRaisesOpError(
"cannot contain fractional components."):
binom.cdf(placeholder).eval(
feed_dict={placeholder: [1.0, 2.5, 1.5]})
binom = binomial.Binomial(total_count=n,
probs=p,
validate_args=False)
binom.prob([1., 2., 3.]).eval()
binom.cdf([1., 2., 3.]).eval()
# Non-integer arguments work.
binom.prob([1.0, 2.5, 1.5]).eval()
binom.cdf([1.0, 2.5, 1.5]).eval()
def testPmfNonIntegerCounts(self):
p = [[0.1, 0.2, 0.7]]
n = [[5.]]
with self.test_session():
# No errors with integer n.
binom = binomial.Binomial(n=n, p=p, validate_args=True)
binom.pmf([2., 3, 2]).eval()
binom.pmf([3., 1, 2]).eval()
# Both equality and integer checking fail.
with self.assertRaisesOpError("Condition x == y.*"):
binom.pmf([1.0, 2.5, 1.5]).eval()
binom = binomial.Binomial(n=n, p=p, validate_args=False)
binom.pmf([1., 2., 3.]).eval()
# Non-integer arguments work.
binom.pmf([1.0, 2.5, 1.5]).eval()
def testPProperty(self):
p = [[0.1, 0.2, 0.7]]
with self.test_session():
binom = binomial.Binomial(n=3., p=p)
self.assertEqual((1, 3), binom.p.get_shape())
self.assertEqual((1, 3), binom.logits.get_shape())
self.assertAllClose(p, binom.p.eval())
def testNProperty(self):
p = [[0.1, 0.2, 0.7], [0.2, 0.3, 0.5]]
n = [[3.], [4]]
with self.test_session():
binom = binomial.Binomial(n=n, p=p)
self.assertEqual((2, 1), binom.n.get_shape())
self.assertAllClose(n, binom.n.eval())
def testPmfPStretchedInBroadcastWhenSameRank(self):
with self.test_session():
p = [[0.1, 0.9]]
counts = [[1., 2.]]
pmf = binomial.Binomial(total_count=3., probs=p).prob(counts)
self.assertAllClose(stats.binom.pmf(counts, n=3., p=p), pmf.eval())
self.assertEqual((1, 2), pmf.get_shape())
def testNProperty(self):
p = [[0.1, 0.2, 0.7], [0.2, 0.3, 0.5]]
n = [[3.], [4]]
with self.cached_session():
binom = binomial.Binomial(total_count=n, probs=p)
self.assertEqual((2, 1), binom.total_count.get_shape())
self.assertAllClose(n, binom.total_count.eval())
def testLogitsProperty(self):
logits = [[0., 9., -0.5]]
with self.cached_session():
binom = binomial.Binomial(total_count=3., logits=logits)
self.assertEqual((1, 3), binom.probs.get_shape())
self.assertEqual((1, 3), binom.logits.get_shape())
self.assertAllClose(logits, binom.logits.eval())
def testPProperty(self):
p = [[0.1, 0.2, 0.7]]
with self.cached_session():
binom = binomial.Binomial(total_count=3., probs=p)
self.assertEqual((1, 3), binom.probs.get_shape())
self.assertEqual((1, 3), binom.logits.get_shape())
self.assertAllClose(p, binom.probs.eval())
def testPmfPStretchedInBroadcastWhenLowerRank(self):
with self.test_session():
p = [0.1, 0.4]
counts = [[1.], [0.]]
pmf = binomial.Binomial(total_count=1., probs=p).prob(counts)
self.assertAllClose([[0.1, 0.4], [0.9, 0.6]], pmf.eval())
self.assertEqual((2, 2), pmf.get_shape())
def testSimpleShapes(self):
with self.cached_session():
p = np.float32(np.random.beta(1, 1))
binom = binomial.Binomial(total_count=1., probs=p)
self.assertAllEqual([], binom.event_shape_tensor().eval())
self.assertAllEqual([], binom.batch_shape_tensor().eval())
self.assertEqual(tensor_shape.TensorShape([]), binom.event_shape)
self.assertEqual(tensor_shape.TensorShape([]), binom.batch_shape)
def testBinomialMode(self):
with self.cached_session():
n = 5.
p = [0.1, 0.2, 0.7]
binom = binomial.Binomial(total_count=n, probs=p)
expected_modes = [0., 1, 4]
self.assertEqual((3, ), binom.mode().get_shape())
self.assertAllClose(expected_modes, binom.mode().eval())
def testBinomialVariance(self):
with self.cached_session():
n = 5.
p = [0.1, 0.2, 0.7]
binom = binomial.Binomial(total_count=n, probs=p)
expected_variances = stats.binom.var(n, p)
self.assertEqual((3, ), binom.variance().get_shape())
self.assertAllClose(expected_variances, binom.variance().eval())
def testBinomialMean(self):
with self.test_session():
n = 5.
p = [0.1, 0.2, 0.7]
binom = binomial.Binomial(total_count=n, probs=p)
expected_means = stats.binom.mean(n, p)
self.assertEqual((3,), binom.mean().get_shape())
self.assertAllClose(expected_means, binom.mean().eval())
def testPmfBothZeroBatches(self):
with self.test_session():
# Both zero-batches. No broadcast
p = 0.5
counts = 1.
pmf = binomial.Binomial(total_count=1., probs=p).prob(counts)
self.assertAllClose(0.5, pmf.eval())
self.assertEqual((), pmf.get_shape())
def testPmfBothZeroBatchesNontrivialN(self):
with self.test_session():
# Both zero-batches. No broadcast
p = 0.1
counts = 3.
binom = binomial.Binomial(total_count=5., probs=p)
pmf = binom.prob(counts)
self.assertAllClose(stats.binom.pmf(counts, n=5., p=p), pmf.eval())
self.assertEqual((), pmf.get_shape())
def testComplexShapes(self):
with self.cached_session():
p = np.random.beta(1, 1, size=(3, 2)).astype(np.float32)
n = [[3., 2], [4, 5], [6, 7]]
binom = binomial.Binomial(total_count=n, probs=p)
self.assertAllEqual([], binom.event_shape_tensor().eval())
self.assertAllEqual([3, 2], binom.batch_shape_tensor().eval())
self.assertEqual(tensor_shape.TensorShape([]), binom.event_shape)
self.assertEqual(tensor_shape.TensorShape([3, 2]),
binom.batch_shape)
def testSimpleShapes(self):
with self.test_session():
p = np.float32(np.random.beta(1, 1))
binom = binomial.Binomial(n=1., p=p)
self.assertAllEqual([], binom.event_shape().eval())
self.assertAllEqual([], binom.batch_shape().eval())
self.assertEqual(tensor_shape.TensorShape([]),
binom.get_event_shape())
self.assertEqual(tensor_shape.TensorShape([]),
binom.get_batch_shape())
def testPmfNandCountsAgree(self):
p = [[0.1, 0.2, 0.7]]
n = [[5.]]
with self.test_session():
binom = binomial.Binomial(total_count=n, probs=p, validate_args=True)
binom.prob([2., 3, 2]).eval()
binom.prob([3., 1, 2]).eval()
with self.assertRaisesOpError("Condition x >= 0.*"):
binom.prob([-1., 4, 2]).eval()
with self.assertRaisesOpError("Condition x <= y.*"):
binom.prob([7., 3, 0]).eval()
def testComplexShapes(self):
with self.test_session():
p = np.random.beta(1, 1, size=(3, 2)).astype(np.float32)
n = [[3., 2], [4, 5], [6, 7]]
binom = binomial.Binomial(n=n, p=p)
self.assertAllEqual([], binom.event_shape().eval())
self.assertAllEqual([3, 2], binom.batch_shape().eval())
self.assertEqual(tensor_shape.TensorShape([]),
binom.get_event_shape())
self.assertEqual(tensor_shape.TensorShape([3, 2]),
binom.get_batch_shape())
def testBinomialMultipleMode(self):
with self.cached_session():
n = 9.
p = [0.1, 0.2, 0.7]
binom = binomial.Binomial(total_count=n, probs=p)
# For the case where (n + 1) * p is an integer, the modes are:
# (n + 1) * p and (n + 1) * p - 1. In this case, we get back
# the larger of the two modes.
expected_modes = [1., 2, 7]
self.assertEqual((3, ), binom.mode().get_shape())
self.assertAllClose(expected_modes, binom.mode().eval())
def testPmfAndCdfBothZeroBatches(self):
with self.cached_session():
# Both zero-batches. No broadcast
p = 0.5
counts = 1.
binom = binomial.Binomial(total_count=1., probs=p)
pmf = binom.prob(counts)
cdf = binom.cdf(counts)
self.assertAllClose(0.5, pmf.eval())
self.assertAllClose(stats.binom.cdf(counts, n=1, p=p), cdf.eval())
self.assertEqual((), pmf.get_shape())
self.assertEqual((), cdf.get_shape())