# Copyright 2022-2023 XProbe Inc.
# derived from copyright 1999-2021 Alibaba Group Holding 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.
from numbers import Number
import numpy as np
from ... import opcodes as OperandDef
from ...core import ENTITY_TYPE
from ...serialization.serializables import AnyField, KeyField
from ..array_utils import as_same_device, device
from ..core import Tensor
from ..datasource import tensor as astensor
from ..utils import broadcast_shape
from .core import TensorElementWise, TensorOperand, filter_inputs
class TensorClip(TensorOperand, TensorElementWise):
_op_type_ = OperandDef.CLIP
_a = KeyField("a")
_a_min = AnyField("a_min")
_a_max = AnyField("a_max")
_out = KeyField("out")
def __init__(self, a=None, a_min=None, a_max=None, out=None, **kw):
super().__init__(_a=a, _a_min=a_min, _a_max=a_max, _out=out, **kw)
@property
def a(self):
return self._a
@property
def a_min(self):
return self._a_min
@property
def a_max(self):
return self._a_max
@property
def out(self):
return getattr(self, "_out", None)
def _set_inputs(self, inputs):
super()._set_inputs(inputs)
inputs_iter = iter(self._inputs)
self._a = next(inputs_iter)
if isinstance(self._a_min, ENTITY_TYPE):
self._a_min = next(inputs_iter)
if isinstance(self._a_max, ENTITY_TYPE):
self._a_max = next(inputs_iter)
if getattr(self, "_out", None) is not None:
self._out = next(inputs_iter)
def __call__(self, a, a_min, a_max, out=None):
a = astensor(a)
tensors = [a]
sparse = a.issparse()
if isinstance(a_min, Number):
if a_min > 0:
sparse = False
a_min_dtype = np.array(a_min).dtype
elif a_min is not None:
a_min = astensor(a_min)
tensors.append(a_min)
if not a_min.issparse():
sparse = False
a_min_dtype = a_min.dtype
else:
a_min_dtype = None
self._a_min = a_min
if isinstance(a_max, Number):
if a_max < 0:
sparse = False
a_max_dtype = np.array(a_max).dtype
elif a_max is not None:
a_max = astensor(a_max)
tensors.append(a_max)
if not a_max.issparse():
sparse = False
a_max_dtype = a_max.dtype
else:
a_max_dtype = None
self._a_max = a_max
if out is not None:
if isinstance(out, Tensor):
self._out = out
else:
raise TypeError(f"out should be Tensor object, got {type(out)} instead")
dtypes = [dt for dt in [a.dtype, a_min_dtype, a_max_dtype] if dt is not None]
dtype = np.result_type(*dtypes)
# check broadcast
shape = broadcast_shape(*[t.shape for t in tensors])
setattr(self, "sparse", sparse)
inputs = filter_inputs([a, a_min, a_max, out])
t = self.new_tensor(inputs, shape)
if out is None:
setattr(self, "dtype", dtype)
return t
# if `out` is specified, use out's dtype and shape
out_shape, out_dtype = out.shape, out.dtype
if t.shape != out_shape:
t = self.new_tensor(inputs, out_shape)
setattr(self, "dtype", out_dtype)
out.data = t.data
return out
@classmethod
def execute(cls, ctx, op):
inputs, device_id, xp = as_same_device(
[ctx[c.key] for c in op.inputs], device=op.device, ret_extra=True
)
inputs_iter = iter(inputs)
a = next(inputs_iter)
a_min = (
next(inputs_iter) if isinstance(op.a_min, type(op.outputs[0])) else op.a_min
)
a_max = (
next(inputs_iter) if isinstance(op.a_max, type(op.outputs[0])) else op.a_max
)
out = next(inputs_iter).copy() if op.out is not None else None
with device(device_id):
kw = {}
if out is not None:
kw["out"] = out
ctx[op.outputs[0].key] = xp.clip(a, a_min, a_max, **kw)
[docs]def clip(a, a_min, a_max, out=None):
"""
Clip (limit) the values in a tensor.
Given an interval, values outside the interval are clipped to
the interval edges. For example, if an interval of ``[0, 1]``
is specified, values smaller than 0 become 0, and values larger
than 1 become 1.
Parameters
----------
a : array_like
Tensor containing elements to clip.
a_min : scalar or array_like or `None`
Minimum value. If `None`, clipping is not performed on lower
interval edge. Not more than one of `a_min` and `a_max` may be
`None`.
a_max : scalar or array_like or `None`
Maximum value. If `None`, clipping is not performed on upper
interval edge. Not more than one of `a_min` and `a_max` may be
`None`. If `a_min` or `a_max` are array_like, then the three
arrays will be broadcasted to match their shapes.
out : Tensor, optional
The results will be placed in this tensor. It may be the input
array for in-place clipping. `out` must be of the right shape
to hold the output. Its type is preserved.
Returns
-------
clipped_array : Tensor
An tensor with the elements of `a`, but where values
< `a_min` are replaced with `a_min`, and those > `a_max`
with `a_max`.
Examples
--------
>>> import mars.tensor as mt
>>> a = mt.arange(10)
>>> mt.clip(a, 1, 8).execute()
array([1, 1, 2, 3, 4, 5, 6, 7, 8, 8])
>>> a.execute()
array([0, 1, 2, 3, 4, 5, 6, 7, 8, 9])
>>> mt.clip(a, 3, 6, out=a).execute()
array([3, 3, 3, 3, 4, 5, 6, 6, 6, 6])
>>> a = mt.arange(10)
>>> a.execute()
array([0, 1, 2, 3, 4, 5, 6, 7, 8, 9])
>>> mt.clip(a, [3, 4, 1, 1, 1, 4, 4, 4, 4, 4], 8).execute()
array([3, 4, 2, 3, 4, 5, 6, 7, 8, 8])
"""
op = TensorClip(a=a, a_min=a_min, a_max=a_max, out=out)
return op(a, a_min, a_max, out=out)