# Copyright 2020 Huawei Technologies Co., Ltd
#
# 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.
# ============================================================================
"""Scalar Affine Bijector"""
from mindspore.ops import operations as P
from ..distribution._utils.custom_ops import log_generic
from .bijector import Bijector
[docs]class ScalarAffine(Bijector):
"""
Scalar Affine Bijector.
This Bijector performs the operation:
.. math::
Y = a * X + b
where a is the scale factor and b is the shift factor.
Args:
scale (float, list, numpy.ndarray, Tensor): The scale factor. Default: 1.0.
shift (float, list, numpy.ndarray, Tensor): The shift factor. Default: 0.0.
name (str): The name of the bijector. Default: 'ScalarAffine'.
Supported Platforms:
``Ascend`` ``GPU``
Note:
The dtype of `shift` and `scale` must be float.
If `shift`, `scale` are passed in as numpy.ndarray or tensor, they have to have
the same dtype otherwise an error will be raised.
Raises:
TypeError: When the dtype of `shift` or `scale` is not float,
and when the dtype of `shift` and `scale` is not same.
Examples:
>>> import mindspore
>>> import mindspore.nn as nn
>>> from mindspore import Tensor
>>>
>>> # To initialize a ScalarAffine bijector of scale 1.0 and shift 2.
>>> scalaraffine = nn.probability.bijector.ScalarAffine(1.0, 2.0)
>>> value = Tensor([1, 2, 3], dtype=mindspore.float32)
>>> ans1 = scalaraffine.forward(value)
>>> print(ans1.shape)
(3,)
>>> ans2 = scalaraffine.inverse(value)
>>> print(ans2.shape)
(3,)
>>> ans3 = scalaraffine.forward_log_jacobian(value)
>>> print(ans3.shape)
()
>>> ans4 = scalaraffine.inverse_log_jacobian(value)
>>> print(ans4.shape)
()
"""
def __init__(self,
scale=1.0,
shift=0.0,
name='ScalarAffine'):
"""
Constructor of ScalarAffine Bijector.
"""
param = dict(locals())
param['param_dict'] = {'scale': scale, 'shift': shift}
super(ScalarAffine, self).__init__(
is_constant_jacobian=True,
is_injective=True,
name=name,
dtype=None,
param=param)
self._scale = self._add_parameter(scale, 'scale')
self._shift = self._add_parameter(shift, 'shift')
self.abs = P.Abs()
self.oneslike = P.OnesLike()
self.dtypeop = P.DType()
self.cast = P.Cast()
self.log = log_generic
@property
def scale(self):
return self._scale
@property
def shift(self):
return self._shift
def extend_repr(self):
if self.is_scalar_batch:
str_info = f'scale = {self.scale}, shift = {self.shift}'
else:
str_info = f'batch_shape = {self.batch_shape}'
return str_info
def _forward(self, x):
r"""
.. math::
f(x) = a * x + b
"""
x = self._check_value_dtype(x)
scale_local = self.cast_param_by_value(x, self.scale)
shift_local = self.cast_param_by_value(x, self.shift)
forward_v = scale_local * x + shift_local * self.oneslike(x)
return forward_v
def _inverse(self, y):
r"""
.. math::
f(y) = \frac{y - b}{a}
"""
y = self._check_value_dtype(y)
scale_local = self.cast_param_by_value(y, self.scale)
shift_local = self.cast_param_by_value(y, self.shift)
inverse_v = (y - shift_local) / scale_local
return inverse_v
def _forward_log_jacobian(self, x):
r"""
.. math::
f(x) = a * x + b
f'(x) = a
\log(f'(x)) = \log(a)
"""
x = self._check_value_dtype(x)
scale_local = self.cast_param_by_value(x, self.scale)
forward_log_j = self.log(self.abs(scale_local))
return forward_log_j
def _inverse_log_jacobian(self, y):
r"""
.. math::
f(y) = \frac{(y - b)}{a}
f'(x) = \frac{1.0}{a}
\log(f'(x)) = - \log(a)
"""
y = self._check_value_dtype(y)
scale_local = self.cast_param_by_value(y, self.scale)
inverse_log_j = -1. * self.log(self.abs(scale_local))
return inverse_log_j