Revision fc7e5a6dae3d8c1365e99ac5d64d8f696496e69f authored by vdutor on 11 October 2019, 17:54:54 UTC, committed by vdutor on 11 October 2019, 17:54:54 UTC
1 parent c23c6db
test_expectations.py
# Copyright 2018 the GPflow authors.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
import numpy as np
import pytest
import tensorflow as tf
from numpy.testing import assert_allclose
from tensorflow import convert_to_tensor as ctt
import gpflow
from gpflow import inducing_variables, kernels
from gpflow import mean_functions as mf
from gpflow.config import default_float
from gpflow.expectations import expectation, quadrature_expectation
from gpflow.probability_distributions import (DiagonalGaussian, Gaussian,
MarkovGaussian)
rng = np.random.RandomState(1)
RTOL = 1e-6
num_data = 5
num_ind = 4
D_in = 2
D_out = 2
Xmu = ctt(rng.randn(num_data, D_in))
Xmu_markov = ctt(rng.randn(num_data + 1, D_in)) # (N+1)xD
Xcov = rng.randn(num_data, D_in, D_in)
Xcov = ctt(Xcov @ np.transpose(Xcov, (0, 2, 1)))
Z = rng.randn(num_ind, D_in)
def markov_gauss():
cov_params = rng.randn(num_data + 1, D_in, 2 * D_in) / 2. # (N+1)xDx2D
Xcov = cov_params @ np.transpose(cov_params, (0, 2, 1)) # (N+1)xDxD
Xcross = cov_params[:-1] @ np.transpose(cov_params[1:], (0, 2, 1)) # NxDxD
Xcross = np.concatenate((Xcross, np.zeros((1, D_in, D_in))),
0) # (N+1)xDxD
Xcov = np.stack([Xcov, Xcross]) # 2x(N+1)xDxD
return MarkovGaussian(Xmu_markov, ctt(Xcov))
_means = {
'lin': mf.Linear(A=rng.randn(D_in, D_out), b=rng.randn(D_out)),
'identity': mf.Identity(input_dim=D_in),
'const': mf.Constant(c=rng.randn(D_out)),
'zero': mf.Zero(output_dim=D_out)
}
_distrs = {
'gauss':
Gaussian(Xmu, Xcov),
'dirac_gauss':
Gaussian(Xmu, np.zeros((num_data, D_in, D_in))),
'gauss_diag':
DiagonalGaussian(Xmu, rng.rand(num_data, D_in)),
'dirac_diag':
DiagonalGaussian(Xmu, np.zeros((num_data, D_in))),
'dirac_markov_gauss':
MarkovGaussian(Xmu_markov, np.zeros((2, num_data + 1, D_in, D_in))),
'markov_gauss':
markov_gauss()
}
_kerns = {
'rbf':
kernels.SquaredExponential(variance=rng.rand(), lengthscale=rng.rand() + 1.),
'lin':
kernels.Linear(variance=rng.rand()),
'matern':
kernels.Matern32(variance=rng.rand()),
'rbf_act_dim_0':
kernels.SquaredExponential(variance=rng.rand(),
lengthscale=rng.rand() + 1.,
active_dims=[0]),
'rbf_act_dim_1':
kernels.SquaredExponential(variance=rng.rand(),
lengthscale=rng.rand() + 1.,
active_dims=[1]),
'lin_act_dim_0':
kernels.Linear(variance=rng.rand(), active_dims=[0]),
'lin_act_dim_1':
kernels.Linear(variance=rng.rand(), active_dims=[1]),
'rbf_lin_sum':
kernels.Sum([
kernels.SquaredExponential(variance=rng.rand(), lengthscale=rng.rand() + 1.),
kernels.Linear(variance=rng.rand())
]),
'rbf_lin_sum2':
kernels.Sum([
kernels.Linear(variance=rng.rand()),
kernels.SquaredExponential(variance=rng.rand(), lengthscale=rng.rand() + 1.),
kernels.Linear(variance=rng.rand()),
kernels.SquaredExponential(variance=rng.rand(), lengthscale=rng.rand() + 1.),
]),
'rbf_lin_prod':
kernels.Product([
kernels.SquaredExponential(variance=rng.rand(),
lengthscale=rng.rand() + 1.,
active_dims=[0]),
kernels.Linear(variance=rng.rand(), active_dims=[1])
])
}
def kerns(*args):
return [_kerns[k] for k in args]
def distrs(*args):
return [_distrs[k] for k in args]
def means(*args):
return [_means[k] for k in args]
@pytest.fixture
def inducing_variable():
return inducing_variables.InducingPoints(Z)
def _check(params):
analytic = expectation(*params)
quad = quadrature_expectation(*params)
assert_allclose(analytic, quad, rtol=RTOL)
# =================================== TESTS ===================================
distr_args1 = distrs("gauss")
mean_args = means("lin", "identity", "const", "zero")
kern_args1 = kerns("lin", "rbf", "rbf_lin_sum", "rbf_lin_prod")
kern_args2 = kerns("lin", "rbf", "rbf_lin_sum")
@pytest.mark.parametrize("distribution", distr_args1)
@pytest.mark.parametrize("mean1", mean_args)
@pytest.mark.parametrize("mean2", mean_args)
@pytest.mark.parametrize(
"arg_filter", [lambda p, m1, m2: (p, m1), lambda p, m1, m2: (p, m1, m2)])
def test_mean_function_only_expectations(distribution, mean1, mean2,
arg_filter):
params = arg_filter(distribution, mean1, mean2)
_check(params)
@pytest.mark.parametrize("distribution", distrs("gauss", "gauss_diag"))
@pytest.mark.parametrize("kernel", kern_args1)
@pytest.mark.parametrize("arg_filter", [
lambda p, k, f: (p, k), lambda p, k, f: (p, (k, f)), lambda p, k, f:
(p, (k, f), (k, f))
])
def test_kernel_only_expectations(distribution, kernel, inducing_variable, arg_filter):
params = arg_filter(distribution, kernel, inducing_variable)
_check(params)
@pytest.mark.parametrize("distribution", distr_args1)
@pytest.mark.parametrize("kernel", kerns("rbf", "lin", "matern",
"rbf_lin_sum"))
@pytest.mark.parametrize("mean", mean_args)
@pytest.mark.parametrize(
"arg_filter",
[lambda p, k, f, m: (p, (k, f), m), lambda p, k, f, m: (p, m, (k, f))])
def test_kernel_mean_function_expectations(distribution, kernel, inducing_variable, mean,
arg_filter):
params = arg_filter(distribution, kernel, inducing_variable, mean)
_check(params)
@pytest.mark.parametrize("kernel", kern_args1)
def test_eKdiag_no_uncertainty(kernel):
eKdiag = expectation(_distrs['dirac_diag'], kernel)
Kdiag = kernel(Xmu, full=False)
assert_allclose(eKdiag, Kdiag, rtol=RTOL)
@pytest.mark.parametrize("kernel", kern_args1)
def test_eKxz_no_uncertainty(kernel, inducing_variable):
eKxz = expectation(_distrs['dirac_diag'], (kernel, inducing_variable))
Kxz = kernel(Xmu, Z)
assert_allclose(eKxz, Kxz, rtol=RTOL)
@pytest.mark.parametrize("kernel", kern_args2)
@pytest.mark.parametrize("mean", mean_args)
def test_eMxKxz_no_uncertainty(kernel, inducing_variable, mean):
exKxz = expectation(_distrs['dirac_diag'], mean, (kernel, inducing_variable))
Kxz = kernel(Xmu, Z)
xKxz = expectation(_distrs['dirac_gauss'],
mean)[:, :, None] * Kxz[:, None, :]
assert_allclose(exKxz, xKxz, rtol=RTOL)
@pytest.mark.parametrize("kernel", kern_args1)
def test_eKzxKxz_no_uncertainty(kernel, inducing_variable):
eKzxKxz = expectation(_distrs['dirac_diag'], (kernel, inducing_variable),
(kernel, inducing_variable))
Kxz = kernel(Xmu, Z)
KzxKxz = Kxz[:, :, None] * Kxz[:, None, :]
assert_allclose(eKzxKxz, KzxKxz, rtol=RTOL)
def test_RBF_eKzxKxz_gradient_notNaN():
"""
Ensure that <K_{Z, x} K_{x, Z}>_p(x) is not NaN and correct, when
K_{Z, Z} is zero with finite precision. See pull request #595.
"""
kernel = gpflow.kernels.SquaredExponential(1, lengthscale=0.1)
kernel.variance <<= 2.
p = gpflow.probability_distributions.Gaussian(
tf.constant([[10]], dtype=default_float()),
tf.constant([[[0.1]]], dtype=default_float()))
z = gpflow.inducing_variables.InducingPoints([[-10.], [10.]])
with tf.GradientTape() as tape:
ekz = expectation(p, (kernel, z), (kernel, z))
grad = tape.gradient(ekz, kernel.lengthscale)
assert grad is not None and not np.isnan(grad)
@pytest.mark.parametrize("distribution", distrs("gauss_diag"))
@pytest.mark.parametrize("kern1", kerns("rbf_act_dim_0", "lin_act_dim_0"))
@pytest.mark.parametrize("kern2", kerns("rbf_act_dim_1", "lin_act_dim_1"))
def test_eKzxKxz_separate_dims_simplification(distribution, kern1, kern2,
inducing_variable):
_check((distribution, (kern1, inducing_variable), (kern2, inducing_variable)))
@pytest.mark.parametrize("distribution", distr_args1)
@pytest.mark.parametrize("kern1", kerns("rbf_lin_sum"))
@pytest.mark.parametrize("kern2", kerns("rbf_lin_sum2"))
def test_eKzxKxz_different_sum_kernels(distribution, kern1, kern2, inducing_variable):
_check((distribution, (kern1, inducing_variable), (kern2, inducing_variable)))
@pytest.mark.parametrize("distribution", distr_args1)
@pytest.mark.parametrize("kern1", kerns("rbf_lin_sum2"))
@pytest.mark.parametrize("kern2", kerns("rbf_lin_sum2"))
def test_eKzxKxz_same_vs_different_sum_kernels(distribution, kern1, kern2,
inducing_variable):
# check the result is the same if we pass different objects with the same value
same = expectation(*(distribution, (kern1, inducing_variable), (kern1, inducing_variable)))
different = expectation(*(distribution, (kern1, inducing_variable),
(kern2, inducing_variable)))
assert_allclose(same, different, rtol=RTOL)
@pytest.mark.parametrize("distribution", distrs("markov_gauss"))
@pytest.mark.parametrize("kernel", kern_args2)
@pytest.mark.parametrize("mean", means("identity"))
def test_exKxz_markov(distribution, kernel, mean, inducing_variable):
_check((distribution, (kernel, inducing_variable), mean))
@pytest.mark.parametrize("distribution", distrs("dirac_markov_gauss"))
@pytest.mark.parametrize("kernel", kern_args2)
@pytest.mark.parametrize("mean", means("identity"))
def test_exKxz_markov_no_uncertainty(distribution, kernel, mean, inducing_variable):
exKxz = expectation(distribution, (kernel, inducing_variable), mean)
Kzx = kernel(Xmu_markov[:-1, :], Z) # NxM
xKxz = Kzx[..., None] * Xmu_markov[1:, None, :] # NxMxD
assert_allclose(exKxz, xKxz, rtol=RTOL)
@pytest.mark.parametrize("kernel", kerns("rbf"))
@pytest.mark.parametrize("distribution",
distrs("gauss", "gauss_diag", "markov_gauss"))
def test_cov_shape_inference(distribution, kernel, inducing_variable):
gauss_tuple = (distribution.mu, distribution.cov)
_check((gauss_tuple, (kernel, inducing_variable)))
if isinstance(distribution, MarkovGaussian):
_check((gauss_tuple, None, (kernel, inducing_variable)))
Computing file changes ...