def testValidateArgs(self):
with self.assertRaises(tf.errors.InvalidArgumentError):
k = psd_kernels.RationalQuadratic(amplitude=-1.,
length_scale=1.,
scale_mixture_rate=1.,
validate_args=True)
self.evaluate(k.amplitude)
with self.assertRaises(tf.errors.InvalidArgumentError):
k = psd_kernels.RationalQuadratic(amplitude=1.,
length_scale=-1.,
scale_mixture_rate=1.,
validate_args=True)
self.evaluate(k.length_scale)
with self.assertRaises(tf.errors.InvalidArgumentError):
k = psd_kernels.RationalQuadratic(amplitude=1.,
length_scale=1.,
scale_mixture_rate=-1.,
validate_args=True)
self.evaluate(k.scale_mixture_rate)
# But `None`'s are ok
k = psd_kernels.RationalQuadratic(amplitude=None,
length_scale=None,
scale_mixture_rate=None,
validate_args=True)
self.evaluate(k.apply([1.], [1.]))
def testShapesAreCorrect(self):
k = psd_kernels.RationalQuadratic(amplitude=1., length_scale=1.)
x = np.ones([4, 3], np.float32)
y = np.ones([5, 3], np.float32)
self.assertAllEqual([4, 5], k.matrix(x, y).shape)
self.assertAllEqual([2, 4, 5],
k.matrix(tf.stack([x] * 2),
tf.stack([y] * 2)).shape)
k = psd_kernels.RationalQuadratic(
amplitude=np.ones([2, 1, 1], np.float32),
length_scale=np.ones([1, 3, 1], np.float32),
scale_mixture_rate=np.ones([2, 1, 1, 1], np.float32))
self.assertAllEqual(
[2, 2, 3, 2, 4, 5],
#`-----' | `--'
# | | `- matrix shape
# | `- from input batch shapes
# `- from broadcasting kernel params
k.matrix(
tf.stack([x] * 2), # shape [2, 4, 3]
tf.stack([y] * 2) # shape [2, 5, 3]
).shape)
def testBatchShape(self):
amplitude = np.random.uniform(2, 3., size=[3, 1, 2]).astype(np.float32)
length_scale = np.random.uniform(2, 3., size=[1, 3, 1]).astype(np.float32)
scale_mixture_rate = np.random.uniform(
2, 3., size=[3, 1, 1]).astype(np.float32)
k = tfpk.RationalQuadratic(
amplitude, length_scale, scale_mixture_rate)
self.assertAllEqual(tf.TensorShape([3, 3, 2]), k.batch_shape)
self.assertAllEqual([3, 3, 2], self.evaluate(k.batch_shape_tensor()))
def testShapesAreCorrect(self):
k = psd_kernels.RationalQuadratic(amplitude=1., length_scale=1.)
x = np.ones([4, 3], np.float32)
y = np.ones([5, 3], np.float32)
self.assertAllEqual(k.matrix(x, y).shape, [4, 5])
self.assertAllEqual(
k.matrix(tf.stack([x] * 2), tf.stack([y] * 2)).shape, [2, 4, 5])
k = psd_kernels.RationalQuadratic(
amplitude=np.ones([2, 1, 1], np.float32),
length_scale=np.ones([1, 3, 1], np.float32),
scale_mixture_rate=np.ones([2, 1, 1, 1], np.float32))
self.assertAllEqual(
k.matrix(
tf.stack([x] * 2), # shape [2, 4, 3]
tf.stack([y] * 2) # shape [2, 5, 3]
).shape,
[2, 2, 3, 2, 4, 5])
def testValuesAreCorrect(self, feature_ndims, dtype, dims):
amplitude = np.array(5., dtype=dtype)
length_scale = np.array(.2, dtype=dtype)
scale_mixture_rate = np.array(3., dtype=dtype)
np.random.seed(42)
k = psd_kernels.RationalQuadratic(amplitude, length_scale,
scale_mixture_rate, feature_ndims)
shape = [dims] * feature_ndims
for _ in range(5):
x = np.random.uniform(-1, 1, size=shape).astype(dtype)
y = np.random.uniform(-1, 1, size=shape).astype(dtype)
self.assertAllClose(
self._rational_quadratic(amplitude, length_scale,
scale_mixture_rate, x, y),
self.evaluate(k.apply(x, y)))
def testNoneScaleMixture(self):
amplitude = 5.
length_scale = .2
np.random.seed(42)
k = tfpk.RationalQuadratic(
amplitude=amplitude,
length_scale=length_scale,
scale_mixture_rate=None)
x = np.random.uniform(-1, 1, size=[5]).astype(np.float32)
y = np.random.uniform(-1, 1, size=[5]).astype(np.float32)
self.assertAllClose(
# Ensure that a None value for scale_mixture_rate has the same semantics
# as scale_mixture_rate=1.
self._rational_quadratic(
amplitude=amplitude,
length_scale=length_scale, scale_mixture_rate=1., x=x, y=y),
self.evaluate(k.apply(x, y)))
def testMismatchedFloatTypesAreBad(self):
with self.assertRaises(TypeError):
psd_kernels.RationalQuadratic(np.float32(1.), np.float64(1.))
