def testDocstringExample(self):
with self.test_session():
exp_gamma_distribution = (
transformed_distribution_lib.TransformedDistribution(
distribution=gamma_lib.Gamma(concentration=1., rate=2.),
bijector=bijectors.Invert(bijectors.Exp())))
self.assertAllEqual(
[], array_ops.shape(exp_gamma_distribution.sample()).eval())
def testGammaModeAllowNanStatsIsFalseWorksWhenAllBatchMembersAreDefined(
self):
alpha_v = np.array([5.5, 3.0, 2.5])
beta_v = np.array([1.0, 4.0, 5.0])
gamma = gamma_lib.Gamma(concentration=alpha_v, rate=beta_v)
expected_modes = (alpha_v - 1) / beta_v
self.assertEqual(gamma.mode().get_shape(), (3, ))
self.assertAllClose(self.evaluate(gamma.mode()), expected_modes)
def testGammaShape(self):
alpha = constant_op.constant([3.0] * 5)
beta = constant_op.constant(11.0)
gamma = gamma_lib.Gamma(concentration=alpha, rate=beta)
self.assertEqual(self.evaluate(gamma.batch_shape_tensor()), (5, ))
self.assertEqual(gamma.batch_shape, tensor_shape.TensorShape([5]))
self.assertAllEqual(self.evaluate(gamma.event_shape_tensor()), [])
self.assertEqual(gamma.event_shape, tensor_shape.TensorShape([]))
def testGammaEntropy(self):
alpha_v = np.array([1.0, 3.0, 2.5])
beta_v = np.array([1.0, 4.0, 5.0])
gamma = gamma_lib.Gamma(concentration=alpha_v, rate=beta_v)
self.assertEqual(gamma.entropy().get_shape(), (3, ))
if not stats:
return
expected_entropy = stats.gamma.entropy(alpha_v, scale=1 / beta_v)
self.assertAllClose(self.evaluate(gamma.entropy()), expected_entropy)
def testGammaModeAllowNanStatsFalseRaisesForUndefinedBatchMembers(self):
# Mode will not be defined for the first entry.
alpha_v = np.array([0.5, 3.0, 2.5])
beta_v = np.array([1.0, 4.0, 5.0])
gamma = gamma_lib.Gamma(concentration=alpha_v,
rate=beta_v,
allow_nan_stats=False)
with self.assertRaisesOpError("x < y"):
self.evaluate(gamma.mode())
def testGammaStd(self):
with self.test_session():
alpha_v = np.array([1.0, 3.0, 2.5])
beta_v = np.array([1.0, 4.0, 5.0])
gamma = gamma_lib.Gamma(concentration=alpha_v, rate=beta_v)
self.assertEqual(gamma.stddev().get_shape(), (3, ))
if not stats:
return
expected_stddev = stats.gamma.std(alpha_v, scale=1. / beta_v)
self.assertAllClose(self.evaluate(gamma.stddev()), expected_stddev)
def test_docstring_example_gamma(self):
with self.test_session() as sess:
num_draws = int(1e5)
concentration_p = constant_op.constant(1.)
concentration_q = constant_op.constant(2.)
p = gamma_lib.Gamma(concentration=concentration_p, rate=1.)
q = gamma_lib.Gamma(concentration=concentration_q, rate=3.)
approx_kl_gamma_gamma = monte_carlo_lib.expectation(
f=lambda x: p.log_prob(x) - q.log_prob(x),
samples=p.sample(num_draws, seed=42),
log_prob=p.log_prob,
use_reparametrization=(p.reparameterization_type
== distribution_lib.FULLY_REPARAMETERIZED))
exact_kl_gamma_gamma = kullback_leibler.kl_divergence(p, q)
[exact_kl_gamma_gamma_, approx_kl_gamma_gamma_] = sess.run([
exact_kl_gamma_gamma, approx_kl_gamma_gamma])
self.assertEqual(
False,
p.reparameterization_type == distribution_lib.FULLY_REPARAMETERIZED)
self.assertAllClose(exact_kl_gamma_gamma_, approx_kl_gamma_gamma_,
rtol=0.01, atol=0.)
# Compare gradients. (Not present in `docstring`.)
gradp = lambda fp: gradients_impl.gradients(fp, concentration_p)[0]
gradq = lambda fq: gradients_impl.gradients(fq, concentration_q)[0]
[
gradp_exact_kl_gamma_gamma_,
gradq_exact_kl_gamma_gamma_,
gradp_approx_kl_gamma_gamma_,
gradq_approx_kl_gamma_gamma_,
] = sess.run([
gradp(exact_kl_gamma_gamma),
gradq(exact_kl_gamma_gamma),
gradp(approx_kl_gamma_gamma),
gradq(approx_kl_gamma_gamma),
])
# Notice that variance (i.e., `rtol`) is higher when using score-trick.
self.assertAllClose(gradp_exact_kl_gamma_gamma_,
gradp_approx_kl_gamma_gamma_,
rtol=0.05, atol=0.)
self.assertAllClose(gradq_exact_kl_gamma_gamma_,
gradq_approx_kl_gamma_gamma_,
rtol=0.03, atol=0.)
def testGammaVariance(self):
with self.test_session():
alpha_v = np.array([1.0, 3.0, 2.5])
beta_v = np.array([1.0, 4.0, 5.0])
gamma = gamma_lib.Gamma(concentration=alpha_v, rate=beta_v)
self.assertEqual(gamma.variance().get_shape(), (3, ))
if not stats:
return
expected_variances = stats.gamma.var(alpha_v, scale=1 / beta_v)
self.assertAllClose(gamma.variance().eval(), expected_variances)
def testGammaFullyReparameterized(self):
alpha = constant_op.constant(4.0)
beta = constant_op.constant(3.0)
with backprop.GradientTape() as tape:
tape.watch(alpha)
tape.watch(beta)
gamma = gamma_lib.Gamma(concentration=alpha, rate=beta)
samples = gamma.sample(100)
grad_alpha, grad_beta = tape.gradient(samples, [alpha, beta])
self.assertIsNotNone(grad_alpha)
self.assertIsNotNone(grad_beta)
def testGammaModeAllowNanStatsIsTrueReturnsNaNforUndefinedBatchMembers(
self):
# Mode will not be defined for the first entry.
alpha_v = np.array([0.5, 3.0, 2.5])
beta_v = np.array([1.0, 4.0, 5.0])
gamma = gamma_lib.Gamma(concentration=alpha_v,
rate=beta_v,
allow_nan_stats=True)
expected_modes = (alpha_v - 1) / beta_v
expected_modes[0] = np.nan
self.assertEqual(gamma.mode().get_shape(), (3, ))
self.assertAllClose(self.evaluate(gamma.mode()), expected_modes)
def testGammaCDF(self):
batch_size = 6
alpha = constant_op.constant([2.0] * batch_size)
beta = constant_op.constant([3.0] * batch_size)
alpha_v = 2.0
beta_v = 3.0
x = np.array([2.5, 2.5, 4.0, 0.1, 1.0, 2.0], dtype=np.float32)
gamma = gamma_lib.Gamma(concentration=alpha, rate=beta)
cdf = gamma.cdf(x)
self.assertEqual(cdf.get_shape(), (6, ))
if not stats:
return
expected_cdf = stats.gamma.cdf(x, alpha_v, scale=1 / beta_v)
self.assertAllClose(self.evaluate(cdf), expected_cdf)
def testGammaLogPDF(self):
with self.test_session():
batch_size = 6
alpha = constant_op.constant([2.0] * batch_size)
beta = constant_op.constant([3.0] * batch_size)
alpha_v = 2.0
beta_v = 3.0
x = np.array([2.5, 2.5, 4.0, 0.1, 1.0, 2.0], dtype=np.float32)
gamma = gamma_lib.Gamma(concentration=alpha, rate=beta)
log_pdf = gamma.log_prob(x)
self.assertEqual(log_pdf.get_shape(), (6, ))
pdf = gamma.prob(x)
self.assertEqual(pdf.get_shape(), (6, ))
if not stats:
return
expected_log_pdf = stats.gamma.logpdf(x, alpha_v, scale=1 / beta_v)
self.assertAllClose(self.evaluate(log_pdf), expected_log_pdf)
self.assertAllClose(self.evaluate(pdf), np.exp(expected_log_pdf))
def testGammaLogPDFMultidimensional(self):
batch_size = 6
alpha = constant_op.constant([[2.0, 4.0]] * batch_size)
beta = constant_op.constant([[3.0, 4.0]] * batch_size)
alpha_v = np.array([2.0, 4.0])
beta_v = np.array([3.0, 4.0])
x = np.array([[2.5, 2.5, 4.0, 0.1, 1.0, 2.0]], dtype=np.float32).T
gamma = gamma_lib.Gamma(concentration=alpha, rate=beta)
log_pdf = gamma.log_prob(x)
log_pdf_values = self.evaluate(log_pdf)
self.assertEqual(log_pdf.get_shape(), (6, 2))
pdf = gamma.prob(x)
pdf_values = self.evaluate(pdf)
self.assertEqual(pdf.get_shape(), (6, 2))
if not stats:
return
expected_log_pdf = stats.gamma.logpdf(x, alpha_v, scale=1 / beta_v)
self.assertAllClose(log_pdf_values, expected_log_pdf)
self.assertAllClose(pdf_values, np.exp(expected_log_pdf))
def testStateParts(self):
with self.test_session(graph=ops.Graph()) as sess:
dist_x = normal_lib.Normal(loc=self.dtype(0), scale=self.dtype(1))
dist_y = independent_lib.Independent(
gamma_lib.Gamma(concentration=self.dtype([1, 2]),
rate=self.dtype([0.5, 0.75])),
reinterpreted_batch_ndims=1)
def target_log_prob(x, y):
return dist_x.log_prob(x) + dist_y.log_prob(y)
x0 = [dist_x.sample(seed=1), dist_y.sample(seed=2)]
samples, _ = hmc.sample_chain(num_results=int(2e3),
target_log_prob_fn=target_log_prob,
current_state=x0,
step_size=0.85,
num_leapfrog_steps=3,
num_burnin_steps=int(250),
seed=49)
actual_means = [math_ops.reduce_mean(s, axis=0) for s in samples]
actual_vars = [_reduce_variance(s, axis=0) for s in samples]
expected_means = [dist_x.mean(), dist_y.mean()]
expected_vars = [dist_x.variance(), dist_y.variance()]
[
actual_means_,
actual_vars_,
expected_means_,
expected_vars_,
] = sess.run([
actual_means,
actual_vars,
expected_means,
expected_vars,
])
self.assertAllClose(expected_means_,
actual_means_,
atol=0.05,
rtol=0.16)
self.assertAllClose(expected_vars_,
actual_vars_,
atol=0.,
rtol=0.25)
def testGammaSample(self):
alpha_v = 4.0
beta_v = 3.0
alpha = constant_op.constant(alpha_v)
beta = constant_op.constant(beta_v)
n = 100000
gamma = gamma_lib.Gamma(concentration=alpha, rate=beta)
samples = gamma.sample(n, seed=137)
sample_values = self.evaluate(samples)
self.assertEqual(samples.get_shape(), (n, ))
self.assertEqual(sample_values.shape, (n, ))
self.assertTrue(self._kstest(alpha_v, beta_v, sample_values))
if not stats:
return
self.assertAllClose(sample_values.mean(),
stats.gamma.mean(alpha_v, scale=1 / beta_v),
atol=.01)
self.assertAllClose(sample_values.var(),
stats.gamma.var(alpha_v, scale=1 / beta_v),
atol=.15)
def testGammaPdfOfSampleMultiDims(self):
gamma = gamma_lib.Gamma(concentration=[7., 11.], rate=[[5.], [6.]])
num = 50000
samples = gamma.sample(num, seed=137)
pdfs = gamma.prob(samples)
sample_vals, pdf_vals = self.evaluate([samples, pdfs])
self.assertEqual(samples.get_shape(), (num, 2, 2))
self.assertEqual(pdfs.get_shape(), (num, 2, 2))
self._assertIntegral(sample_vals[:, 0, 0], pdf_vals[:, 0, 0], err=0.02)
self._assertIntegral(sample_vals[:, 0, 1], pdf_vals[:, 0, 1], err=0.02)
self._assertIntegral(sample_vals[:, 1, 0], pdf_vals[:, 1, 0], err=0.02)
self._assertIntegral(sample_vals[:, 1, 1], pdf_vals[:, 1, 1], err=0.02)
if not stats:
return
self.assertAllClose(stats.gamma.mean([[7., 11.], [7., 11.]],
scale=1 /
np.array([[5., 5.], [6., 6.]])),
sample_vals.mean(axis=0),
atol=.1)
self.assertAllClose(stats.gamma.var([[7., 11.], [7., 11.]],
scale=1 /
np.array([[5., 5.], [6., 6.]])),
sample_vals.var(axis=0),
atol=.1)
