def testMeanCovarianceNoBatch(self):
with self.test_session() as sess:
dims = 3
vdm = vector_diffeomixture_lib.VectorDiffeomixture(
mix_loc=[[0.], [4.]],
mix_scale=[10.],
distribution=normal_lib.Normal(0., 1.),
loc=[
np.float32([-2.]),
None,
],
scale=[
linop_identity_lib.LinearOperatorScaledIdentity(
num_rows=dims,
multiplier=np.float32(1.5),
is_positive_definite=True),
linop_diag_lib.LinearOperatorDiag(
diag=np.linspace(2.5, 3.5, dims, dtype=np.float32),
is_positive_definite=True),
],
validate_args=True)
self.run_test_sample_consistent_mean_covariance(sess.run,
vdm,
rtol=0.02,
cov_rtol=0.06)
def _cholesky_diag(diag_operator):
return linear_operator_diag.LinearOperatorDiag(math_ops.sqrt(
diag_operator.diag),
is_non_singular=True,
is_self_adjoint=True,
is_positive_definite=True,
is_square=True)
def _inverse_diag(diag_operator):
return linear_operator_diag.LinearOperatorDiag(
1. / diag_operator.diag,
is_non_singular=diag_operator.is_non_singular,
is_self_adjoint=diag_operator.is_self_adjoint,
is_positive_definite=diag_operator.is_positive_definite,
is_square=True)
def linop_scale(w, op):
# We assume w > 0. (This assumption only relates to the is_* attributes.)
with ops.name_scope("linop_scale", values=[w]):
# TODO(b/35301104): LinearOperatorComposition doesn't combine operators, so
# special case combinations here. Once it does, this function can be
# replaced by:
# return linop_composition_lib.LinearOperatorComposition([
# scaled_identity(w), op])
def scaled_identity(w):
return linop_identity_lib.LinearOperatorScaledIdentity(
num_rows=op.range_dimension_tensor(),
multiplier=w,
is_non_singular=op.is_non_singular,
is_self_adjoint=op.is_self_adjoint,
is_positive_definite=op.is_positive_definite)
if isinstance(op, linop_identity_lib.LinearOperatorIdentity):
return scaled_identity(w)
if isinstance(op, linop_identity_lib.LinearOperatorScaledIdentity):
return scaled_identity(w * op.multiplier)
if isinstance(op, linop_diag_lib.LinearOperatorDiag):
return linop_diag_lib.LinearOperatorDiag(
diag=w[..., array_ops.newaxis] * op.diag_part(),
is_non_singular=op.is_non_singular,
is_self_adjoint=op.is_self_adjoint,
is_positive_definite=op.is_positive_definite)
if isinstance(op, linop_tril_lib.LinearOperatorLowerTriangular):
return linop_tril_lib.LinearOperatorLowerTriangular(
tril=w[..., array_ops.newaxis, array_ops.newaxis] *
op.to_dense(),
is_non_singular=op.is_non_singular,
is_self_adjoint=op.is_self_adjoint,
is_positive_definite=op.is_positive_definite)
raise NotImplementedError("Unsupported Linop type ({})".format(
type(op).__name__))
def _add(self, op1, op2, operator_name, hints):
return linear_operator_diag.LinearOperatorDiag(
diag=op1.diag_part() + op2.diag_part(),
is_non_singular=hints.is_non_singular,
is_self_adjoint=hints.is_self_adjoint,
is_positive_definite=hints.is_positive_definite,
name=operator_name)
def test_none_loc_static_scale(self):
loc = None
scale = linear_operator_diag.LinearOperatorDiag(np.ones((5, 1, 3)))
batch_shape, event_shape = distribution_util.shapes_from_loc_and_scale(
loc, scale)
self.assertEqual(tensor_shape.TensorShape([5, 1]), batch_shape)
self.assertEqual(tensor_shape.TensorShape([3]), event_shape)
def _set_diag_operators(self, diag_update, is_diag_update_positive):
"""Set attributes self._diag_update and self._diag_operator."""
if diag_update is not None:
self._diag_operator = linear_operator_diag.LinearOperatorDiag(
self._diag_update,
is_positive_definite=is_diag_update_positive)
self._diag_inv_operator = linear_operator_diag.LinearOperatorDiag(
1. / self._diag_update,
is_positive_definite=is_diag_update_positive)
else:
if self.u.get_shape()[-1].value is not None:
r = self.u.get_shape()[-1].value
else:
r = array_ops.shape(self.u)[-1]
self._diag_operator = linear_operator_identity.LinearOperatorIdentity(
num_rows=r, dtype=self.dtype)
self._diag_inv_operator = self._diag_operator
def _matmul_linear_operator_diag(linop_a, linop_b):
return linear_operator_diag.LinearOperatorDiag(
diag=linop_a.diag * linop_b.diag,
is_non_singular=_combined_non_singular_hint(linop_a, linop_b),
is_self_adjoint=_combined_self_adjoint_hint(linop_a, linop_b),
is_positive_definite=_combined_positive_definite_hint(
linop_a, linop_b),
is_square=True)
def test_none_loc_dynamic_scale(self):
loc = None
diag = array_ops.placeholder(dtypes.float64)
scale = linear_operator_diag.LinearOperatorDiag(diag)
with self.test_session() as sess:
batch_shape, event_shape = sess.run(
distribution_util.shapes_from_loc_and_scale(loc, scale),
feed_dict={diag: np.ones((5, 1, 3))})
self.assertAllEqual([5, 1], batch_shape)
self.assertAllEqual([3], event_shape)
def _solve_linear_operator_diag(linop_a, linop_b):
return linear_operator_diag.LinearOperatorDiag(
diag=linop_b.diag / linop_a.diag,
is_non_singular=registrations_util.combined_non_singular_hint(
linop_a, linop_b),
is_self_adjoint=registrations_util.
combined_commuting_self_adjoint_hint(linop_a, linop_b),
is_positive_definite=(
registrations_util.combined_commuting_positive_definite_hint(
linop_a, linop_b)),
is_square=True)
def _adjoint_diag(diag_operator):
diag = diag_operator.diag
if diag.dtype.is_complex:
diag = math_ops.conj(diag)
return linear_operator_diag.LinearOperatorDiag(
diag=diag,
is_non_singular=diag_operator.is_non_singular,
is_self_adjoint=diag_operator.is_self_adjoint,
is_positive_definite=diag_operator.is_positive_definite,
is_square=True)
def _matmul_linear_operator_diag_scaled_identity_left(linop_scaled_identity,
linop_diag):
return linear_operator_diag.LinearOperatorDiag(
diag=linop_diag.diag * linop_scaled_identity.multiplier,
is_non_singular=_combined_non_singular_hint(linop_diag,
linop_scaled_identity),
is_self_adjoint=_combined_self_adjoint_hint(linop_diag,
linop_scaled_identity),
is_positive_definite=_combined_positive_definite_hint(
linop_diag, linop_scaled_identity),
is_square=True)
def _set_diag_operators(self, diag_update, is_diag_update_positive):
"""Set attributes self._diag_update and self._diag_operator."""
if diag_update is not None:
self._diag_operator = linear_operator_diag.LinearOperatorDiag(
self._diag_update, is_positive_definite=is_diag_update_positive)
else:
if tensor_shape.dimension_value(self.u.shape[-1]) is not None:
r = tensor_shape.dimension_value(self.u.shape[-1])
else:
r = array_ops.shape(self.u)[-1]
self._diag_operator = linear_operator_identity.LinearOperatorIdentity(
num_rows=r, dtype=self.dtype)
def _matmul_linear_operator_diag_scaled_identity_right(
linop_diag, linop_scaled_identity):
return linear_operator_diag.LinearOperatorDiag(
diag=linop_diag.diag * linop_scaled_identity.multiplier,
is_non_singular=registrations_util.combined_non_singular_hint(
linop_diag, linop_scaled_identity),
is_self_adjoint=registrations_util.combined_commuting_self_adjoint_hint(
linop_diag, linop_scaled_identity),
is_positive_definite=(
registrations_util.combined_commuting_positive_definite_hint(
linop_diag, linop_scaled_identity)),
is_square=True)
def test_dynamic_loc_static_scale(self):
loc = array_ops.placeholder(dtypes.float64)
diag = constant_op.constant(np.ones((5, 2, 3)))
scale = linear_operator_diag.LinearOperatorDiag(diag)
with self.test_session():
batch_shape, event_shape = distribution_util.shapes_from_loc_and_scale(
loc, scale)
# batch_shape depends on both args, and so is dynamic. Since loc did not
# have static shape, we inferred event shape entirely from scale, and this
# is available statically.
self.assertAllEqual(
[5, 2], batch_shape.eval(feed_dict={loc: np.zeros((2, 3))}))
self.assertAllEqual([3], event_shape)
def testGlobalDispatcherLinearOperators(self):
original_global_dispatchers = dispatch._GLOBAL_DISPATCHERS
try:
TensorTracerOpDispatcher().register()
x = TensorTracer("x")
# To grab the eigenvalues the diag operator just calls convert_to_tensor
# (twice) in this case.
trace = linear_operator_diag.LinearOperatorDiag(x).eigvals()
self.assertEqual(
str(trace),
"convert_to_tensor(convert_to_tensor(x, dtype=None, dtype_hint=None, "
"name=diag))")
# The diagonal tensor addition gets traced even though the linear_operator
# API only uses dispatchable ops instead of directly exposing dispatching.
trace = linear_operator_diag.LinearOperatorDiag(x).add_to_tensor(x)
self.assertIn(
"linalg.set_diag(convert_to_tensor(x, name=x), __operators__.add("
"convert_to_tensor(x, dtype=None, dtype_hint=None, name=diag), "
"linalg.diag_part(convert_to_tensor(x, name=x)), "
"name=", str(trace))
# The dispatch-supporting ops the non-singular check calls out to
# get traced.
trace = linear_operator_diag.LinearOperatorDiag(
x).assert_non_singular()
self.assertIn("debugging.assert_less", str(trace))
self.assertIn(
"message=Singular operator: Diagonal contained zero values.",
str(trace))
finally:
# Clean up.
dispatch._GLOBAL_DISPATCHERS = original_global_dispatchers
def testSampleProbConsistentDynamicQuadrature(self):
with self.test_session() as sess:
qgrid = array_ops.placeholder(dtype=dtypes.float32)
qprobs = array_ops.placeholder(dtype=dtypes.float32)
g, p = np.polynomial.hermite.hermgauss(deg=8)
dims = 4
vdm = vector_diffeomixture_lib.VectorDiffeomixture(
mix_loc=[[0.], [1.]],
mix_scale=[1.],
distribution=normal_lib.Normal(0., 1.),
loc=[
None,
np.float32([2.] * dims),
],
scale=[
linop_identity_lib.LinearOperatorScaledIdentity(
num_rows=dims,
multiplier=np.float32(1.1),
is_positive_definite=True),
linop_diag_lib.LinearOperatorDiag(
diag=np.linspace(2.5, 3.5, dims, dtype=np.float32),
is_positive_definite=True),
],
quadrature_grid_and_probs=(g, p),
validate_args=True)
# Ball centered at component0's mean.
sess_run_fn = lambda x: sess.run(x,
feed_dict={
qgrid: g,
qprobs: p
})
self.run_test_sample_consistent_log_prob(sess_run_fn,
vdm,
radius=2.,
center=0.,
rtol=0.005)
# Larger ball centered at component1's mean.
self.run_test_sample_consistent_log_prob(sess_run_fn,
vdm,
radius=4.,
center=2.,
rtol=0.005)
def testSampleProbConsistentBroadcastMixBatch(self):
with self.test_session() as sess:
dims = 4
vdm = vdm_lib.VectorDiffeomixture(
mix_loc=[[0.], [1.]],
temperature=[1.],
distribution=normal_lib.Normal(0., 1.),
loc=[
None,
np.float32([2.] * dims),
],
scale=[
linop_identity_lib.LinearOperatorScaledIdentity(
num_rows=dims,
multiplier=[np.float32(1.1)],
is_positive_definite=True),
linop_diag_lib.LinearOperatorDiag(
diag=np.stack([
np.linspace(2.5, 3.5, dims, dtype=np.float32),
np.linspace(2.75, 3.25, dims, dtype=np.float32),
]),
is_positive_definite=True),
],
quadrature_size=8,
validate_args=True)
# Ball centered at component0's mean.
self.run_test_sample_consistent_log_prob(sess.run,
vdm,
radius=2.,
center=0.,
rtol=0.01)
# Larger ball centered at component1's mean.
self.run_test_sample_consistent_log_prob(sess.run,
vdm,
radius=4.,
center=2.,
rtol=0.01)
def testMeanCovarianceNoBatchUncenteredNonStandardBase(self):
with self.cached_session() as sess:
dims = 3
vdm = vdm_lib.VectorDiffeomixture(
mix_loc=[[0.], [4.]],
temperature=[0.1],
distribution=normal_lib.Normal(-1., 1.5),
loc=[
np.float32([-2.]),
np.float32([0.]),
],
scale=[
linop_identity_lib.LinearOperatorScaledIdentity(
num_rows=dims,
multiplier=np.float32(1.5),
is_positive_definite=True),
linop_diag_lib.LinearOperatorDiag(
diag=np.linspace(2.5, 3.5, dims, dtype=np.float32),
is_positive_definite=True),
],
quadrature_size=8,
validate_args=True)
self.run_test_sample_consistent_mean_covariance(
sess.run, vdm, num_samples=int(1e6), rtol=0.01, cov_atol=0.025)
def _mean_of_covariance_given_quadrature_component(self, diag_only):
p = self.mixture_distribution.probs
# To compute E[Cov(Z|V)], we'll add matrices within three categories:
# scaled-identity, diagonal, and full. Then we'll combine these at the end.
scale_identity_multiplier = None
diag = None
full = None
for k, aff in enumerate(self.interpolated_affine):
s = aff.scale # Just in case aff.scale has side-effects, we'll call once.
if (s is None or isinstance(
s, linop_identity_lib.LinearOperatorIdentity)):
scale_identity_multiplier = add(scale_identity_multiplier,
p[..., k, array_ops.newaxis])
elif isinstance(s,
linop_identity_lib.LinearOperatorScaledIdentity):
scale_identity_multiplier = add(
scale_identity_multiplier, (p[..., k, array_ops.newaxis] *
math_ops.square(s.multiplier)))
elif isinstance(s, linop_diag_lib.LinearOperatorDiag):
diag = add(diag, (p[..., k, array_ops.newaxis] *
math_ops.square(s.diag_part())))
else:
x = (p[..., k, array_ops.newaxis, array_ops.newaxis] *
s.matmul(s.to_dense(), adjoint_arg=True))
if diag_only:
x = array_ops.matrix_diag_part(x)
full = add(full, x)
# We must now account for the fact that the base distribution might have a
# non-unity variance. Recall that, since X ~ iid Law(X_0),
# `Cov(SX+m) = S Cov(X) S.T = S S.T Diag(Var(X_0))`.
# We can scale by `Var(X)` (vs `Cov(X)`) since X corresponds to `d` iid
# samples from a scalar-event distribution.
v = self.distribution.variance()
if scale_identity_multiplier is not None:
scale_identity_multiplier *= v
if diag is not None:
diag *= v[..., array_ops.newaxis]
if full is not None:
full *= v[..., array_ops.newaxis]
if diag_only:
# Apparently we don't need the full matrix, just the diagonal.
r = add(diag, full)
if r is None and scale_identity_multiplier is not None:
ones = array_ops.ones(self.event_shape_tensor(),
dtype=self.dtype)
return scale_identity_multiplier[..., array_ops.newaxis] * ones
return add(r, scale_identity_multiplier)
# `None` indicates we don't know if the result is positive-definite.
is_positive_definite = (True if all(
aff.scale.is_positive_definite
for aff in self.endpoint_affine) else None)
to_add = []
if diag is not None:
to_add.append(
linop_diag_lib.LinearOperatorDiag(
diag=diag, is_positive_definite=is_positive_definite))
if full is not None:
to_add.append(
linop_full_lib.LinearOperatorFullMatrix(
matrix=full, is_positive_definite=is_positive_definite))
if scale_identity_multiplier is not None:
to_add.append(
linop_identity_lib.LinearOperatorScaledIdentity(
num_rows=self.event_shape_tensor()[0],
multiplier=scale_identity_multiplier,
is_positive_definite=is_positive_definite))
return (linop_add_lib.add_operators(to_add)[0].to_dense()
if to_add else None)
def test_static_loc_static_scale_non_matching_event_size_raises(self):
loc = constant_op.constant(np.zeros((2, 4)))
scale = linear_operator_diag.LinearOperatorDiag(np.ones((5, 1, 3)))
with self.assertRaisesRegexp(ValueError, "could not be broadcast"):
distribution_util.shapes_from_loc_and_scale(loc, scale)