def testAsserts(self):
with self.assertRaisesOpError('Argument `scale` must be positive.'):
b = tfb.Weibull(concentration=1., scale=-1., validate_args=True)
self.evaluate(b.forward(-3.))
with self.assertRaisesOpError(
'Argument `concentration` must be positive.'):
b = tfb.Weibull(concentration=-1., scale=1., validate_args=True)
self.evaluate(b.forward(-3.))
def testBijector(self):
with self.test_session():
scale = 5.
concentration = 0.3
bijector = tfb.Weibull(scale=scale,
concentration=concentration,
validate_args=True)
self.assertEqual("weibull", bijector.name)
x = np.array([[[0.], [1.], [14.], [20.], [100.]]],
dtype=np.float32)
# Weibull distribution
weibull_dist = stats.frechet_r(c=concentration, scale=scale)
y = weibull_dist.cdf(x).astype(np.float32)
self.assertAllClose(y, self.evaluate(bijector.forward(x)))
self.assertAllClose(x, self.evaluate(bijector.inverse(y)))
self.assertAllClose(
weibull_dist.logpdf(x),
self.evaluate(
bijector.forward_log_det_jacobian(x, event_ndims=0)))
self.assertAllClose(self.evaluate(
-bijector.inverse_log_det_jacobian(y, event_ndims=0)),
self.evaluate(
bijector.forward_log_det_jacobian(
x, event_ndims=0)),
rtol=1e-4,
atol=0.)
def testScalarCongruency(self):
bijector_test_util.assert_scalar_congruency(tfb.Weibull(
scale=20., concentration=0.3),
lower_x=1.,
upper_x=100.,
eval_func=self.evaluate,
rtol=0.05)
def testBijectorConcentration1LogDetJacobianFiniteAtZero(self): # When concentration = 1., forward_log_det_jacobian should be finite at # zero. scale = np.logspace(0.1, 10., num=20).astype(np.float32) bijector = tfb.Weibull(scale, concentration=1.) fldj = self.evaluate(bijector.forward_log_det_jacobian(0., event_ndims=0)) self.assertAllEqual(np.ones_like(fldj, dtype=np.bool), np.isfinite(fldj))
def testBijectiveAndFinite(self):
bijector = tfb.Weibull(scale=20., concentration=2., validate_args=True)
x = np.linspace(1., 8., num=10).astype(np.float32)
y = np.linspace(
-np.expm1(-1 / 400.),
-np.expm1(-16), num=10).astype(np.float32)
bijector_test_util.assert_bijective_and_finite(
bijector, x, y, eval_func=self.evaluate, event_ndims=0, rtol=1e-3)
def testVariableAssertsScale(self):
concentration = tf.Variable(1.)
scale = tf.Variable(1.)
b = tfb.Weibull(
concentration=concentration, scale=scale, validate_args=True)
# Use identities so that static asserts don't catch the error earlier
# and raise 'Forward transformation input must be at least 0'
minus_1 = tf.identity(tf.convert_to_tensor(-1.))
minus_3 = tf.identity(tf.convert_to_tensor(-3.))
self.evaluate([concentration.initializer, scale.initializer])
with self.assertRaisesOpError('Argument `scale` must be positive.'):
with tf.control_dependencies([scale.assign(minus_1)]):
self.evaluate(b.forward(minus_3))
def testVariableAsserts(self):
concentration = tf.Variable(1.)
scale = tf.Variable(1.)
b = tfb.Weibull(concentration=concentration,
scale=scale,
validate_args=True)
self.evaluate([concentration.initializer, scale.initializer])
with self.assertRaisesOpError('Argument `scale` must be positive.'):
with tf.control_dependencies([scale.assign(-1.)]):
self.evaluate(b.forward(-3.))
with self.assertRaisesOpError(
'Argument `concentration` must be positive.'):
with tf.control_dependencies(
[scale.assign(1.), concentration.assign(-1.)]):
self.evaluate(b.forward(-3.))
def testScalarCongruency(self):
with self.test_session():
assert_scalar_congruency(tfb.Weibull(scale=20., concentration=0.3),
lower_x=1.,
upper_x=100.,
rtol=0.02)
