# Copyright 2022-2023 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.
# ============================================================================
"""Operators for function."""
from __future__ import absolute_import
import builtins
import operator
import numbers
import numpy as np
import mindspore as ms
import mindspore.common.dtype as mstype
from mindspore.ops import operations as P
from mindspore.ops import functional as F
from mindspore.ops.primitive import constexpr
from mindspore.ops.primitive import _primexpr
import mindspore.ops as ops
from mindspore.ops.operations._inner_ops import DynamicBroadcastTo
from mindspore.ops.operations._sequence_ops import TupleToTensor
from mindspore.ops.composite.multitype_ops import _constexpr_utils as const_utils
from mindspore.ops.operations._sequence_ops import TensorToList
from mindspore.ops.auto_generate import OnesLikeExt, ZerosLikeExt, FillScalar, FillTensor, Arange, Chunk, UniqueDim, \
Unique2, SortExt, NonZero, NonZeroExt, Scatter, ScatterValue
from mindspore.ops.auto_generate.gen_ops_prim import SplitTensor
from mindspore.ops.auto_generate.gen_ops_prim import SplitWithSize, RepeatInterleaveInt, RepeatInterleaveTensor
from mindspore.ops.auto_generate.pyboost_inner_prim import _PyboostSearchSortedPrim
from mindspore.ops.operations.array_ops import (
UniqueConsecutive,
MatrixDiagV3,
MatrixDiagPartV3,
MatrixSetDiagV3,
Fills,
Col2Im,
ScatterNdMax,
ScatterNdMul,
IndexFill,
AffineGrid,
Im2Col,
Expand,
Lstsq,
Mvlgamma,
Tril,
Argmax,
ArgMaxWithValue,
ArgMinWithValue
)
from mindspore.common import Tensor
from mindspore.ops._primitive_cache import _get_cache_prim
from mindspore import _checkparam as validator
from mindspore._c_expression import Tensor as Tensor_
from mindspore.ops._utils.utils import ms_arrange
from mindspore.ops.auto_generate import cat, range, scatter_nd, deepcopy, masked_fill, diagonal, expand_dims, \
flip, transpose, triu, unsorted_segment_sum, diag, gather, gather_d, gather_nd, reshape, masked_select, \
broadcast_to, strided_slice, ones, zeros, max_, min_, select, zero_
from mindspore.ops.auto_generate import tensor_scatter_elements as tensor_scatter_elements_ext
from mindspore.ops.auto_generate.gen_ops_prim import scatter_add_ext_op, slice_ext_op, gather_d_op
from mindspore.ops.operations.manually_defined import tile, rank, scalar_cast
from mindspore.ops.auto_generate.pyboost_inner_prim import _PyboostOneHotExtPrim, tril_ext_impl
arg_max_with_value_ = ArgMaxWithValue()
arg_min_with_value_ = ArgMinWithValue()
batch_to_space_nd_v2_ = P.BatchToSpaceNDV2()
cast_ = P.Cast()
diag_ = P.Diag()
dynamic_broadcast_to_ = DynamicBroadcastTo()
eye_ = P.Eye()
fills_ = Fills()
fillv2_ = P.FillV2()
flatten_ = P.Flatten()
gather_ = P.Gather()
gather_d_ = P.GatherD()
gather_nd_ = P.GatherNd()
ger_ = P.Ger()
index_fill_ = IndexFill()
lstsq_ = Lstsq()
matrix_band_part_ = P.array_ops.MatrixBandPart()
ones_ = P.Ones()
population_count_ = P.PopulationCount()
range_ = P.Range()
rank_ = P.Rank()
reduce_max_ = P.ReduceMax()
reduce_min_ = P.ReduceMin()
reshape_ = P.Reshape()
scalar_to_tensor_ = P.ScalarToTensor()
scatter_add_ = P.ScatterAdd()
scatter_div_ = P.ScatterDiv()
scatter_max_ = P.ScatterMax()
scatter_min_ = P.ScatterMin()
scatter_mul_ = P.ScatterMul()
scatter_nd_ = P.ScatterNd()
scatter_update_ = P.ScatterUpdate()
search_sorted_ = _PyboostSearchSortedPrim()
shape_ = P.Shape()
split_tensor = SplitTensor()
split_with_size = SplitWithSize()
size_ = P.Size()
tensor_scatter_add_ = P.TensorScatterAdd()
tensor_scatter_div_ = P.TensorScatterDiv()
tensor_scatter_max_ = P.TensorScatterMax()
tensor_scatter_min_ = P.TensorScatterMin()
tensor_scatter_mul_ = P.TensorScatterMul()
tensor_scatter_sub_ = P.TensorScatterSub()
tensor_select_ = P.Select()
tensor_shape_ = P.TensorShape()
tensor_slice = P.Slice()
tile_ = P.Tile()
transpose_ = P.Transpose()
tuple_to_array_ = P.TupleToArray()
tuple_to_tensor_ = TupleToTensor()
unique_ = P.Unique()
unsorted_segment_max_ = P.UnsortedSegmentMax()
unsorted_segment_min_ = P.UnsortedSegmentMin()
unsorted_segment_prod_ = P.UnsortedSegmentProd()
unsorted_segment_sum_ = P.UnsortedSegmentSum()
ones_like_ = P.OnesLike()
one_hot_ext_impl = _PyboostOneHotExtPrim()
zeros_like_ = P.ZerosLike()
ones_like_ext_ = OnesLikeExt()
zeros_like_ext_ = ZerosLikeExt()
fill_scalar_ = FillScalar()
fill_tensor_ = FillTensor()
sort_ext_ = SortExt()
scatter_ = Scatter()
scatter_value_ = ScatterValue()
arange_ = Arange()
chunk_ = Chunk()
repeat_interleave_int_ = RepeatInterleaveInt()
repeat_interleave_tensor_ = RepeatInterleaveTensor()
unique_dim_ = UniqueDim()
unique2_ = Unique2()
non_zero_ = NonZero()
non_zero_ext_ = NonZeroExt()
@_primexpr
def get_x_shape(x_shape):
if ops.is_sequence_shape_unknown(x_shape):
return (-2,)
if ops.is_sequence_value_unknown(x_shape):
return (-1,)
s = 1
for i in x_shape:
s = s * i
return (s,)
@constexpr
def _check_attr_dtype(param_name, input_dtype, allow_dtypes, cls_name):
validator.check_value_type(param_name, input_dtype, allow_dtypes, cls_name)
check_flatten_order_const = constexpr(validator.check_flatten_order)
##############################
# Tensor Creation Functions.
##############################
def _cast_type(x, to_type):
"""cast input to the specified type or cast input to tensor"""
if isinstance(x, Tensor):
x = cast_(x, to_type)
else:
x = scalar_to_tensor_(x, to_type)
return x
def _get_type(x):
"""get the dtype of input"""
if isinstance(x, Tensor):
return x.dtype
return ops.typeof(x)
def _get_max_type(start, end, step):
"""get max input type with `level`"""
valid_dtypes = [mstype.int32, mstype.float32, mstype.int64, mstype.float64]
arg_map = [start, end, step]
arg_type_map = [str(_get_type(i)) for i in arg_map]
for arg_value in arg_map:
if not (isinstance(arg_value, (float, int))
or (isinstance(arg_value, Tensor) and arg_value.dtype in valid_dtypes)):
raise TypeError(
f"For arange, the input type must be int or float or a TensorScalar in {valid_dtypes},"
f" but got {_get_type(arg_value)}")
type_map = {'Float64': '3', 'Float32': '2', "<class 'float'>": '2', 'Int64': '1', "<class 'int'>": '1',
'Int32': '0'}
type_map_reverse = {'3': mstype.float64,
'2': mstype.float32, '1': mstype.int64, '0': mstype.int32}
type_level = [type_map.get(i) for i in arg_type_map]
max_level = builtins.max(type_level)
return type_map_reverse.get(max_level)
[docs]def arange(start=0, end=None, step=1, *, dtype=None):
r"""
Creates a sequence of numbers that begins at `start` and extends by increments of
`step` up to but not including `end`.
Args:
start (Union[float, int, Tensor], optional): The start of the interval.
If Tensor, the shape must be :math:`()` . Default: ``0`` .
end (Union[float, int, Tensor], optional): The end of the interval, exclusive.
If Tensor, the shape must be :math:`()`.
Default: ``None`` . If ``None`` , it defaults to the value of `start`, and 0 is used as the starting value.
step (Union[float, int, Tensor], optional): Number that increments `start`.
If Tensor, the shape must be :math:`()`. Default: ``1`` .
Keyword Args:
dtype (mindspore.dtype, optional): The required data type of returned Tensor. Default: ``None`` .
When `dtype` is not specified or ``None``:
If `start`, `end`, and `step` are all integers, the dtype of output is int64,
If `start`, `end`, and `step` contain at least one floating-point number, the dtype of output is float32.
Returns:
A 1-D Tensor, with the same type as the inputs.
Raises:
TypeError: If `start`, `end` or `step` is not an int or a float or a TensorScalar(Special Tensor with shape ())
in valid dtypes.
ValueError: If `step` = 0.
ValueError: If `start` >= `end` when `step` > 0.
ValueError: If `start` <= `end` when `step` < 0.
Supported Platforms:
``Ascend`` ``GPU`` ``CPU``
Examples:
>>> import mindspore as ms
>>> from mindspore import Tensor, ops
>>> output = ops.arange(1, 6)
>>> print(output)
[1 2 3 4 5]
>>> print(output.dtype)
Int64
>>> output = ops.arange(0, 3, 1.2)
>>> print(output)
[0. 1.2 2.4]
>>> print(output.dtype)
Float32
>>> output = ops.arange(7, 1, -2)
>>> print(output)
[7 5 3]
>>> print(output.dtype)
Int64
>>> output = ops.arange(ms.Tensor(12.0, dtype=ms.float64), 2, ms.Tensor(-1.0, dtype=ms.float32))
>>> print(output)
[12. 11. 10. 9. 8. 7. 6. 5. 4. 3.]
>>> print(output.dtype)
Float32
"""
if end is None:
start, end = 0, start
max_type = _get_max_type(start, end, step)
start = _cast_type(start, max_type)
end = _cast_type(end, max_type)
step = _cast_type(step, max_type)
if start.shape != () or end.shape != () or step.shape != ():
raise ValueError(f"For arange, the input args must be a TensorScalar,"
f" but got start shape:{start.shape}, end shape:{end.shape}, step shape:{step.shape}")
data = range_(start, end, step)
if dtype is not None:
data = cast_(data, dtype)
return data
[docs]def arange_ext(start=0, end=None, step=1, *, dtype=None):
r"""
Creates a sequence of numbers that begins at `start` and extends by increments of
`step` up to but not including `end`.
Args:
start (Union[float, int], optional): The start of the interval. Default: ``0`` .
end (Union[float, int], optional): The end of the interval, exclusive.
Default: ``None`` . If ``None`` , it defaults to the value of `start`, and 0 is used as the starting value.
step (Union[float, int], optional): The step size with which the array element increments. Default: ``1`` .
Keyword Args:
dtype (mindspore.dtype, optional): The required data type of returned Tensor. Default: ``None`` .
When `dtype` is not specified or ``None``:
If `start`, `end`, and `step` are all integers, the dtype of output is int64,
If `start`, `end`, and `step` contain at least one floating-point number, the dtype of output is float32.
Returns:
A 1-D Tensor, cast to `dtype` if provided, may potentially lose precision due to casting.
Raises:
TypeError: If `start`, `end` or `step` are not of type int or float.
ValueError: If `step` = 0.
ValueError: If `start` >= `end` when `step` > 0.
ValueError: If `start` <= `end` when `step` < 0.
Supported Platforms:
``Ascend``
Examples:
>>> import mindspore as ms
>>> from mindspore import Tensor, ops
>>> output = ops.arange_ext(1, 6)
>>> print(output)
[1 2 3 4 5]
>>> print(output.dtype)
Int64
>>> output = ops.arange_ext(0, 3, 1.2)
>>> print(output)
[0. 1.2 2.4]
>>> print(output.dtype)
Float32
>>> output = ops.arange_ext(7, 1, -2)
>>> print(output)
[7 5 3]
>>> print(output.dtype)
Int64
>>> output = ops.arange_ext(12, 2, -1, dtype=ms.bfloat16)
>>> print(output)
[12. 11. 10. 9. 8. 7. 6. 5. 4. 3.]
>>> print(output.dtype)
BFloat16
"""
if end is None:
start, end = 0, start
return arange_(start, end, step, dtype)
[docs]def concat(tensors, axis=0):
"""
Alias for :func:`mindspore.ops.cat()`.
Tutorial Examples:
- `Tensor - Tensor Operation <https://mindspore.cn/tutorials/en/r2.4.1/beginner/tensor.html#tensor-operation>`_
- `Vision Transformer Image Classification - Building ViT as a whole
<https://mindspore.cn/tutorials/en/r2.4.1/cv/vit.html#building-vit-as-a-whole>`_
- `Sentiment Classification Implemented by RNN - Dense
<https://mindspore.cn/tutorials/en/r2.4.1/nlp/sentiment_analysis.html#dense>`_
"""
return cat(tensors, axis)
[docs]def eye(n, m=None, dtype=None):
"""
Creates a tensor with ones on the diagonal and zeros in the rest.
Note:
The data type of returned tensor can be float16, float32, int8, int16, int32, int64, uint8
or bool on Ascend platforms.
Args:
n (int): The number of rows of returned tensor. Constant value only.
m (int, optional): The number of columns of returned tensor. Constant value only.
Default: ``None`` , if ``None`` , the number of columns is as the same as n.
dtype (mindspore.dtype, optional): MindSpore's dtype, the data type of the returned tensor.
The data type can be bool or Number.
Default: ``None`` , the data type of the returned tensor is mindspore.float32.
Returns:
Tensor, a tensor with ones on the diagonal and the rest of elements are zero. The shape of `output` depends on
the user's Inputs `n` and `m`. And the data type depends on Inputs `dtype`.
Raises:
TypeError: If `m` or `n` is not an int.
ValueError: If `m` or `n` is less than 0.
Supported Platforms:
``Ascend`` ``GPU`` ``CPU``
Examples:
>>> import mindspore
>>> from mindspore import ops
>>> output = ops.eye(2, 2, mindspore.int32)
>>> print(output)
[[1 0]
[0 1]]
>>> print(output.dtype)
Int32
>>> output = ops.eye(1, 2, mindspore.float32)
>>> print(output)
[[1. 0.]]
>>> print(output.dtype)
Float32
>>> output = ops.eye(2, dtype=mindspore.int32)
>>> print(output)
[[1 0]
[0 1]]
>>> print(output.dtype)
Int32
>>> output = ops.eye(2)
>>> print(output)
[[1. 0.]
[0. 1.]]
>>> print(output.dtype)
Float32
"""
if m is None:
m = n
if dtype is None:
dtype = ms.float32
return eye_(n, m, dtype)
[docs]def hamming_window(window_length, periodic=True, alpha=0.54, beta=0.46, *, dtype=None):
r"""
Returns the Hamming window.
.. math::
w[n]=\alpha − \beta \cos \left( \frac{2 \pi n}{N - 1} \right),
where :math:`N` is the full window size.
Args:
window_length (int): The size of returned window. Must be a non negative integer.
periodic (bool, optional): If True, return a periodic window. If False, return a symmetric window.
Default: ``True`` .
alpha (float, optional): The coefficient α. Default: ``0.54`` .
beta (float, optional): The coefficient β. Default: ``0.46`` .
Keyword Args:
dtype (mindspore.dtype, optional): The output window data type. Default: ``None`` .
Returns:
Tensor, a 1-D tensor of size (window_length) containing the window.
Raises:
TypeError: If `window_length` is a negative integer.
TypeError: If `periodic` is not bool.
Supported Platforms:
``Ascend`` ``GPU`` ``CPU``
Examples:
>>> from mindspore import ops
>>> print(ops.hamming_window(6, False))
[0.08 0.39785218 0.91214782 0.91214782 0.39785218 0.08]
"""
if not isinstance(window_length, int):
raise TypeError(f"For array function 'hamming_window', 'window_length' must be int, but got"
f" {type(window_length)}.")
if window_length < 0:
raise ValueError(
f"For array function 'hamming_window', 'window_length' must be non negative number.")
if not isinstance(periodic, bool):
raise TypeError(
f"For array function 'hamming_window', 'periodic' must be bool, but got {type(periodic)}.")
if not isinstance(alpha, float):
raise TypeError(
f"For array function 'hamming_window', 'alpha' must be float, but got {type(alpha)}.")
if not isinstance(beta, float):
raise TypeError(
f"For array function 'hamming_window', 'beta' must be float, but got {type(beta)}.")
if window_length <= 1:
return Tensor(np.ones(window_length))
if dtype is not None and dtype not in mstype.float_type:
raise TypeError(
f"For array function 'hamming_window', 'dtype' must be floating point dtypes, but got {dtype}.")
dtype = mstype.float32 if dtype is None else dtype
op = _get_cache_prim(P.HammingWindow)(periodic, alpha, beta, dtype)
length = Tensor(np.array([window_length]).astype(np.int32))
out = op(length)
return out
[docs]def where(condition, input, other):
r"""
Selects elements from `input` or `other` based on `condition` and returns a tensor.
.. math::
output_i = \begin{cases} input_i,\quad &if\ condition_i \\ other_i,\quad &otherwise \end{cases}
Args:
condition (Tensor[bool]): If True, yield `input`, otherwise yield `other`.
input (Union[Tensor, Scalar]): When `condition` is True, values to select from.
other (Union[Tensor, Scalar]): When `condition` is False, values to select from.
Returns:
Tensor, elements are selected from `input` and `other`.
Raises:
TypeError: If `condition` is not a Tensor.
TypeError: If both `input` and `other` are scalars.
ValueError: If `condition`, `input` and `other` can not broadcast to each other.
Supported Platforms:
``Ascend`` ``GPU`` ``CPU``
Examples:
>>> import numpy as np
>>> from mindspore import Tensor, ops
>>> from mindspore import dtype as mstype
>>> a = Tensor(np.arange(4).reshape((2, 2)), mstype.float32)
>>> b = Tensor(np.ones((2, 2)), mstype.float32)
>>> condition = a < 3
>>> output = ops.where(condition, a, b)
>>> print(output)
[[0. 1.]
[2. 1.]]
"""
return tensor_select_(condition, input, other)
[docs]def reverse(x, axis):
"""
:func:`mindspore.ops.reverse` will be deprecated in the future.
Please use :func:`mindspore.ops.flip` instead.
"""
return flip(x, axis)
[docs]def ravel(input):
"""
Expand the multidimensional Tensor into 1D along the 0 axis direction.
Args:
input (Tensor): A tensor to be flattened.
Returns:
Tensor, a 1-D tensor, containing the same elements of the input.
Raises:
TypeError: If argument `input` is not Tensor.
Supported Platforms:
``Ascend`` ``GPU`` ``CPU``
Examples:
>>> import numpy as np
>>> from mindspore import Tensor, ops
>>> x = Tensor(np.array([[0, 1], [2, 1]]).astype(np.float32))
>>> output = ops.ravel(x)
>>> print(output)
[0. 1. 2. 1.]
>>> print(output.shape)
(4,)
"""
return ops.reshape(input, (-1,))
[docs]def matrix_band_part(x, lower, upper):
r"""
Copy a tensor setting everything outside a central band in each innermost matrix to zero.
.. warning::
This is an experimental API that is subject to change or deletion.
Args:
x (Tensor): Input tensor. :math:`(*, m, n)` where :math:`*` means, any number of additional dimensions.
lower (Union[int, Tensor]): Number of subdiagonals to keep. The data type must be int32 or int64.
If negative, keep entire lower triangle.
upper (Union[int, Tensor]): Number of superdiagonals to keep. The data type must be int32 or int64.
If negative, keep entire upper triangle.
Returns:
Tensor, has the same type and shape as `x`.
Raises:
TypeError: If `x` is not a Tensor.
TypeError: If dtype of `x` is not valid.
TypeError: If `lower` is neither a number nor a Tensor.
TypeError: If `upper` is neither a number nor a Tensor.
TypeError: If dtype of `lower` is neither int32 nor int64.
TypeError: If dtype of `upper` is neither int32 nor int64.
ValueError: If the shape of `x` is not greater than or equal to 2D.
ValueError: If the shape of `lower` is not equal to 0D.
ValueError: If the shape of `upper` is not equal to 0D.
Supported Platforms:
``Ascend`` ``GPU`` ``CPU``
Examples:
>>> import numpy as np
>>> from mindspore import Tensor, ops
>>> x = Tensor(np.ones([2, 4, 4]).astype(np.float32))
>>> output = ops.matrix_band_part(x, 2, 1)
>>> print(output)
[[[1. 1. 0. 0.]
[1. 1. 1. 0.]
[1. 1. 1. 1.]
[0. 1. 1. 1.]]
[[1. 1. 0. 0.]
[1. 1. 1. 0.]
[1. 1. 1. 1.]
[0. 1. 1. 1.]]]
"""
return matrix_band_part_(x, lower, upper)
[docs]def padding(x, pad_dim_size=8):
r"""
Extends the last dimension of the input tensor from 1 to pad_dim_size, by filling with 0.
Args:
x (Tensor): The shape of tensor is :math:`(x_1, x_2, ..., x_R)`. The rank of `x` must be at least 2.
The last dimension of `x` must be 1. The data type is Number.
pad_dim_size (int): The value of the last dimension of `x` to be extended, which must be positive.
Default: ``8`` .
Returns:
Tensor, has the same type and shape as input shape value.
Raises:
TypeError: If `pad_dim_size` is not an int.
ValueError: If `pad_dim_size` is less than 1.
ValueError: If last dim of `x` is not equal to 1.
Supported Platforms:
``Ascend`` ``GPU`` ``CPU``
Examples:
>>> import mindspore
>>> import numpy as np
>>> from mindspore import Tensor, ops
>>> x = Tensor(np.array([[8], [10]]), mindspore.float32)
>>> pad_dim_size = 4
>>> output = ops.padding(x, pad_dim_size)
>>> print(output)
[[ 8. 0. 0. 0.]
[10. 0. 0. 0.]]
"""
padding_ = _get_cache_prim(P.array_ops.Padding)(pad_dim_size)
return padding_(x)
@constexpr
def _check_axis_type(axis, type_int=True, type_tuple=True, type_list=True, ops_name="ops"):
"""Check axis argument type."""
if type_int and isinstance(axis, int):
return True
if (type_tuple and isinstance(axis, tuple)) or (type_list and isinstance(axis, list)):
for ax in axis:
if not isinstance(ax, int):
raise TypeError(
f"For {ops_name}, each axis must be integer, but got {type(ax)} in {axis}.")
return True
type_str = ""
if type_int:
type_str += "int, "
if type_tuple:
type_str += "tuple, "
if type_list:
type_str += "list, "
raise TypeError(
f"For {ops_name}, the axis should be {type_str}, but got {type(axis)}.")
[docs]def one_hot(indices, depth, on_value=1, off_value=0, axis=-1):
r"""
Computes a one-hot tensor.
The locations represented by indices in `indices` take value `on_value`, while all
other locations take value `off_value`.
Note:
If the input `indices` has rank `N`, the output will have rank `N+1`.
The new axis is created at dimension `axis`. On Ascend, if `on_value` is int64 dtype, `indices` must be
int64 dtype, and the value for `on_value` and `off_value` can only be 1 and 0.
Args:
indices(Tensor): A tensor of indices. Tensor of shape :math:`(X_0, \ldots, X_n)`.
Data type must be int32 or int64.
depth(int): A scalar defining the depth of the one-hot dimension.
on_value(Union[Tensor, int, float], optional): A value to fill in output when `indices[j] = i`.
Data type must be int32, int64, float16 or float32. Default: ``1`` .
off_value(Union[Tensor, int, float], optional): A value to fill in output when `indices[j] != i`.
Has the same data type as `on_value`. Default: ``0`` .
axis(int, optional): Position to insert the value. e.g. If shape of `self` is :math:`(N, C)`, and `axis` is -1,
the output shape will be :math:`(N, C, depth)`, If `axis` is 0,
the output shape will be :math:`(depth, N, C)`.
Default: ``-1`` .
Returns:
Tensor, one-hot tensor. Tensor of shape :math:`(X_0, \ldots, X_{axis}, \text{depth} ,X_{axis+1}, \ldots, X_n)`,
and it has the same data type as `on_value`.
Raises:
TypeError: If `axis` or `depth` is not an int.
TypeError: If dtype of `indices` is not int32 or int64.
TypeError: If dtype of `on_value` is not int32, int64, float16 or float32.
TypeError: If `indices`, `on_value` or `off_value` is not a Tensor.
ValueError: If `axis` is not in range [-1, ndim].
ValueError: If `depth` is less than 0.
Supported Platforms:
``Ascend`` ``GPU`` ``CPU``
Examples:
>>> import mindspore
>>> import numpy as np
>>> from mindspore import Tensor, ops
>>> indices = Tensor(np.array([0, 1, 2]), mindspore.int32)
>>> depth, on_value, off_value = 3, Tensor(1.0, mindspore.float32), Tensor(0.0, mindspore.float32)
>>> output = ops.one_hot(indices, depth, on_value, off_value, axis=-1)
>>> print(output)
[[1. 0. 0.]
[0. 1. 0.]
[0. 0. 1.]]
"""
if not isinstance(on_value, Tensor):
on_value = Tensor(on_value)
if not isinstance(off_value, Tensor):
off_value = Tensor(off_value)
onehot = _get_cache_prim(P.OneHot)(axis)
return onehot(indices, depth, on_value, off_value)
[docs]def fill(type, shape, value): # pylint: disable=redefined-outer-name
"""
Create a Tensor of the specified shape and fill it with the specified value.
Args:
type (mindspore.dtype): The specified type of output tensor. The data type only supports
`bool_ <https://www.mindspore.cn/docs/en/r2.4.1/api_python/mindspore/mindspore.dtype.html>`_ and
`number <https://www.mindspore.cn/docs/en/r2.4.1/api_python/mindspore/mindspore.dtype.html>`_ .
shape (Union(Tensor, tuple[int])): The specified shape of output tensor.
value (Union(Tensor, number.Number, bool)): Value to fill the returned tensor.
Returns:
Tensor.
Raises:
TypeError: If `shape` is not a tuple or a tensor.
Supported Platforms:
``Ascend`` ``GPU`` ``CPU``
Examples:
>>> import mindspore
>>> from mindspore import ops
>>> output = ops.fill(mindspore.float32, (2, 2), 1)
>>> print(output)
[[1. 1.]
[1. 1.]]
>>> output = ops.fill(mindspore.float32, (3, 3), 0)
>>> print(output)
[[0. 0. 0.]
[0. 0. 0.]
[0. 0. 0.]]
"""
value = cast_(value, type)
return fillv2_(shape, value)
[docs]def full(size, fill_value, *, dtype=None): # pylint: disable=redefined-outer-name
"""
Create a Tensor of the specified shape and fill it with the specified value.
Args:
size (Union(tuple[int], list[int])): The specified shape of output tensor.
fill_value (number.Number): Value to fill the returned tensor. Complex numbers are not supported for now.
Keyword Args:
dtype (mindspore.dtype): The specified type of output tensor. `bool_` and `number` are supported, for details,
please refer to :class:`mindspore.dtype` . Default: ``None`` .
Returns:
Tensor.
Raises:
TypeError: If `size` is not a tuple or list.
ValueError: The element in `size` is less than 0.
Supported Platforms:
``Ascend`` ``GPU`` ``CPU``
Examples:
>>> from mindspore import ops
>>> output = ops.full((2, 2), 1)
>>> print(output)
[[1. 1.]
[1. 1.]]
>>> output = ops.full((3, 3), 0)
>>> print(output)
[[0. 0. 0.]
[0. 0. 0.]
[0. 0. 0.]]
"""
if not isinstance(size, (list, tuple)):
raise TypeError(
f"For 'ops.full', 'size' must be a tuple or list of ints, but got {type(size)}.")
if dtype is None:
dtype = mstype.int64
if dtype not in mstype.all_types:
raise TypeError(
f"For 'ops.full', 'dtype' must be mindspore.type, but got {dtype}.")
if isinstance(size, list):
size = tuple(size)
return ops.fill(dtype, size, fill_value)
[docs]def full_ext(size, fill_value, *, dtype=None): # pylint: disable=redefined-outer-name
"""
Create a Tensor of the specified shape and fill it with the specified value.
Args:
size (Union(tuple[int], list[int])): The specified shape of output tensor.
fill_value (Union(number.Number, Tensor)): Value to fill the returned tensor. It can be a Scalar number, a 0-D
Tensor, or a 1-D Tensor with only one element.
Keyword Args:
dtype (mindspore.dtype): The specified type of output tensor. `bool_` and `number` are supported, for details,
please refer to :class:`mindspore.dtype` . Default: ``None`` .
Returns:
Tensor.
Raises:
TypeError: If `size` is not a tuple or list.
ValueError: The element in `size` is less than 0.
Supported Platforms:
``Ascend`` ``GPU`` ``CPU``
Examples:
>>> from mindspore import ops
>>> output = ops.full_ext((2, 2), 1)
>>> print(output)
[[1. 1.]
[1. 1.]]
>>> output = ops.full_ext((3, 3), 0)
>>> print(output)
[[0. 0. 0.]
[0. 0. 0.]
[0. 0. 0.]]
"""
return fill_scalar_(size, fill_value, dtype)
[docs]def full_like(input, fill_value, *, dtype=None):
"""
Return a Tensor of the same shape as `input` and filled with `fill_value`.
Args:
input (Tensor): input Tensor and the output Tensor have the same shape as `input`.
fill_value (Number): Value to fill the returned Tensor. Complex numbers are not supported for now.
Keyword Args:
dtype (mindspore.dtype, optional): The specified type of output tensor. `bool_` and `number` are supported,
for details, please refer to :class:`mindspore.dtype` . Default: ``None`` .
Returns:
Tensor.
Raises:
TypeError: If `input` is not a Tensor.
Supported Platforms:
``Ascend`` ``GPU`` ``CPU``
Examples:
>>> import mindspore
>>> from mindspore import Tensor, ops
>>> input = Tensor([[0, 1], [2, 1]], dtype=mindspore.int32)
>>> output = ops.full_like(input, 1)
>>> print(output)
[[1. 1.]
[1. 1.]]
>>> input = Tensor([[0, 1, 1], [2, 1, 2], [1, 3, 4]], dtype=mindspore.int32)
>>> output = ops.full_like(input, 0)
>>> print(output)
[[0. 0. 0.]
[0. 0. 0.]
[0. 0. 0.]]
"""
if not isinstance(input, Tensor):
raise TypeError(
f"For ops.full_like, the argument 'x' must be tensor, but got {type(input)}")
if dtype is None:
dtype = input.dtype
return full(input.shape, fill_value, dtype=dtype)
[docs]def chunk(input, chunks, axis=0):
"""
Cut the input Tensor into `chunks` sub-tensors along the specified axis.
Note:
This function may return less than the specified number of chunks!
Args:
input (Tensor): A Tensor to be cut.
chunks (int): Number of sub-tensors to cut.
axis (int, optional): Specify the dimensions that you want to split. Default: ``0`` .
Returns:
A tuple of sub-tensors.
Raises:
TypeError: If argument `input` is not Tensor.
TypeError: The sum of `chunks` is not int.
TypeError: If argument `axis` is not int.
ValueError: If argument `axis` is out of range of :math:`[-input.ndim, input.ndim)` .
ValueError: If argument `chunks` is not positive number.
Supported Platforms:
``Ascend`` ``GPU`` ``CPU``
Examples:
>>> import numpy as np
>>> from mindspore import ops, Tensor
>>> input_x = np.arange(9).astype("float32")
>>> output = ops.chunk(Tensor(input_x), 3)
>>> print(output)
(Tensor(shape=[3], dtype=Float32, value= [ 0.00000000e+00, 1.00000000e+00, 2.00000000e+00]),
Tensor(shape=[3], dtype=Float32, value= [ 3.00000000e+00, 4.00000000e+00, 5.00000000e+00]),
Tensor(shape=[3], dtype=Float32, value= [ 6.00000000e+00, 7.00000000e+00, 8.00000000e+00]))
"""
if not isinstance(input, Tensor):
raise TypeError(
f'For ops.chunk parameter `input` must be Tensor, but got {type(input)}')
_check_axis_type(axis, True, False, False, "ops.chunk")
arr_axis = _canonicalize_axis(axis, input.ndim)
if not isinstance(chunks, int):
raise TypeError(
f"For ops.chunk type of argument `chunks` should be integer, but got {type(chunks)}")
if chunks <= 0:
raise ValueError(
f"For ops.chunk parameter 'chunks' must be greater than 0, but got {chunks}")
arr_shape = input.shape
length_along_dim = arr_shape[arr_axis]
if length_along_dim == 0:
res = _get_cache_prim(P.Split)(arr_axis)(input)
elif chunks > length_along_dim:
res = _get_cache_prim(P.Split)(arr_axis, length_along_dim)(input)
elif length_along_dim % chunks == 0:
res = _get_cache_prim(P.Split)(arr_axis, chunks)(input)
else:
block_size = int(np.ceil(length_along_dim / chunks))
true_chunks = int(length_along_dim // block_size)
length1 = true_chunks * block_size
length2 = length_along_dim - length1
start1 = _list_comprehensions(rank_(input), 0, True)
size1 = _tuple_setitem(arr_shape, arr_axis, length1)
start2 = _tuple_setitem(start1, arr_axis, length1)
size2 = _tuple_setitem(arr_shape, arr_axis, length2)
res = _get_cache_prim(P.Split)(arr_axis, true_chunks)(
tensor_slice(input, start1, size1))
if length2:
res += _get_cache_prim(P.Split)(arr_axis,
1)(tensor_slice(input, start2, size2))
return res
def chunk_ext(input, chunks, dim=0):
"""
Cut the input Tensor into `chunks` sub-tensors along the specified axis.
Note:
This function may return less than the specified number of chunks!
.. warning::
This is an experimental API that is subject to change or deletion.
Args:
input (Tensor): A Tensor to be cut.
chunks (int): Number of sub-tensors to cut.
dim (int, optional): Specify the dimensions that you want to split. Default: ``0`` .
Returns:
A tuple of sub-tensors.
Raises:
TypeError: If argument `input` is not Tensor.
TypeError: The sum of `chunks` is not int.
TypeError: If argument `dim` is not int.
ValueError: If argument `dim` is out of range of :math:`[-input.ndim, input.ndim)` .
ValueError: If argument `chunks` is not positive number.
Supported Platforms:
``Ascend``
Examples:
>>> import numpy as np
>>> import mindspore
>>> from mindspore import Tensor
>>> input_x = np.arange(9).astype("float32")
>>> output = mindspore.mint.chunk(Tensor(input_x), 3)
>>> print(output)
(Tensor(shape=[3], dtype=Float32, value= [ 0.00000000e+00, 1.00000000e+00, 2.00000000e+00]),
Tensor(shape=[3], dtype=Float32, value= [ 3.00000000e+00, 4.00000000e+00, 5.00000000e+00]),
Tensor(shape=[3], dtype=Float32, value= [ 6.00000000e+00, 7.00000000e+00, 8.00000000e+00]))
"""
return chunk_(input, chunks, dim)
def fills(x, value):
"""
`fills` is deprecated, please use `ops.fill` instead.
"""
if isinstance(value, float):
value_ = value
elif isinstance(value, int):
value_ = float(value)
elif isinstance(value, Tensor):
if value.ndim != 0:
raise ValueError(f"For 'ops.fills', if the argument 'value' is a tensor, the number of its dimension"
f" should be 0, but got {value.ndim}")
value_ = value.astype(mstype.float32)
else:
raise TypeError(f"For 'ops.fills', the type of argument 'value' should be int, float or Tensor,"
f" but got {type(value)}")
return fills_(x, value_)
[docs]def ones_like(input, *, dtype=None):
"""
Returns a Tensor with a value of 1 and its shape is the same as the input.
Args:
input (Tensor): Tensor of any dimension.
Keyword Args:
dtype (:class:`mindspore.dtype`, optional): The specified dtype of the output tensor. If `dtype` is ``None`` ,
the dtype of the input tensor will be used. Default: ``None`` .
Returns:
Tensor, has the same shape as `input` but filled with ones.
Raises:
TypeError: If `input` is not a Tensor.
Supported Platforms:
``Ascend`` ``GPU`` ``CPU``
Examples:
>>> import numpy as np
>>> from mindspore import Tensor, ops
>>> x = Tensor(np.array([[0, 1], [2, 1]]).astype(np.int32))
>>> output = ops.ones_like(x)
>>> print(output)
[[1 1]
[1 1]]
"""
output = ones_like_(input)
_dtype = input.dtype if dtype is None else dtype
output = cast_(output, _dtype)
return output
[docs]def zeros_like(input, *, dtype=None):
r"""
Creates a tensor filled with 0, with the same size as input, and the given dtype.
If `dtype = None`, the tensor will have the same dtype as input `input`.
Args:
input (Tensor): Tensor of any dimension.
Keyword Args:
dtype (:class:`mindspore.dtype`, optional): The specified dtype of the output tensor. If `dtype` is ``None`` ,
the dtype of the input tensor will be used. Default: ``None`` .
Returns:
Tensor, filled with 0.
Raises:
TypeError: If dtype is not a MindSpore dtype.
Supported Platforms:
``Ascend`` ``GPU`` ``CPU``
Examples:
>>> import mindspore
>>> import numpy as np
>>> from mindspore import Tensor, ops
>>> x = Tensor(np.arange(4).reshape(2, 2))
>>> output = ops.zeros_like(x, dtype=mindspore.float32)
>>> print(output)
[[0. 0.]
[0. 0.]]
"""
_dtype = input.dtype if dtype is None else dtype
output = zeros_like_(input)
output = cast_(output, _dtype)
return output
[docs]def ones_like_ext(input, *, dtype=None):
"""
Creates a tensor filled with 1, with the same shape as input, and its data type is determined by the given dtype.
If `dtype = None`, the tensor will have the same dtype as input `input`.
Args:
input (Tensor): Tensor of any dimension.
Keyword Args:
dtype (:class:`mindspore.dtype`, optional): The specified dtype of the output tensor. If `dtype` is ``None`` ,
the dtype of the input tensor will be used. Default: ``None`` .
Returns:
Tensor, has the same shape as `input` but filled with ones.
Raises:
TypeError: If `input` is not a Tensor.
Supported Platforms:
``Ascend`` ``GPU`` ``CPU``
Examples:
>>> import numpy as np
>>> from mindspore import Tensor, ops
>>> x = Tensor(np.array([[0, 1], [2, 1]]).astype(np.int32))
>>> output = ops.function.array_func.ones_like_ext(x)
>>> print(output)
[[1 1]
[1 1]]
"""
return ones_like_ext_(input, dtype)
[docs]def zeros_like_ext(input, *, dtype=None):
r"""
Creates a tensor filled with 0, with the same size as input. Its data type is determined by the given dtype.
If `dtype = None`, the tensor will have the same dtype as input `input`.
Args:
input (Tensor): Tensor of any dimension.
Keyword Args:
dtype (:class:`mindspore.dtype`, optional): The specified dtype of the output tensor. If `dtype` is ``None`` ,
the dtype of the input tensor will be used. Default: ``None`` .
Returns:
Tensor, filled with 0.
Raises:
TypeError: If dtype is not a MindSpore dtype.
Supported Platforms:
``Ascend`` ``GPU`` ``CPU``
Examples:
>>> import mindspore
>>> import numpy as np
>>> from mindspore import Tensor, ops
>>> x = Tensor(np.arange(4).reshape(2, 2))
>>> output = ops.function.array_func.zeros_like_ext(x, dtype=mindspore.float32)
>>> print(output)
[[0. 0.]
[0. 0.]]
"""
return zeros_like_ext_(input, dtype)
##############################
# Tensor Operation Functions.
##############################
[docs]def unique(input):
"""
Returns the unique elements of input tensor and also return a tensor containing the index of each value of input
tensor corresponding to the output unique tensor.
The output contains Tensor `y` and Tensor `idx`, the format is probably similar to (`y`, `idx`).
The shape of Tensor `y` and Tensor `idx` is different in most cases, because Tensor `y` will be deduplicated,
and the shape of Tensor `idx` is consistent with the input.
To get the same shape between `idx` and `y`, please ref to :class:`mindspore.ops.UniqueWithPad` operator.
Args:
input (Tensor): The input tensor.
The shape is :math:`(N,*)` where :math:`*` means, any number of additional dimensions.
.. warning::
This is an experimental API that is subject to change or deletion.
Returns:
Tuple, containing Tensor objects (`y`, `idx`), `y` is a tensor with the
same type as `input`, and contains the unique elements in `input`.
`idx` is a tensor containing indices of elements in
the input corresponding to the output tensor, have the same shape with `input`.
Raises:
TypeError: If `input` is not a Tensor.
Supported Platforms:
``Ascend`` ``GPU`` ``CPU``
Examples:
>>> import mindspore
>>> import numpy as np
>>> from mindspore import Tensor, nn
>>> from mindspore import ops
>>> x = Tensor(np.array([1, 2, 5, 2]), mindspore.int32)
>>> output = ops.unique(x)
>>> print(output)
(Tensor(shape=[3], dtype=Int32, value= [1, 2, 5]), Tensor(shape=[4], dtype=Int32, value= [0, 1, 2, 1]))
>>> y = output[0]
>>> print(y)
[1 2 5]
>>> idx = output[1]
>>> print(idx)
[0 1 2 1]
"""
shape_x = input.shape
length_x = get_x_shape(shape_x)
input = reshape_(input, length_x)
y, idx = unique_(input)
idx = reshape_(idx, shape_x)
return y, idx
[docs]def unique_ext(input, sorted=True, return_inverse=False, return_counts=False, dim=None):
"""
Returns the unique elements of input tensor.
when `return_inverse=True`, also return a tensor containing the index of each value of input
tensor corresponding to the output unique tensor.
when `return_counts=True`, also return a tensor containing the number of occurrences for each
unique value or tensor
Args:
input (Tensor): The input tensor.
sorted(bool): Whether to sort the unique elements in ascending order before returning as output.
Default: ``True`` .
return_inverse(bool): Whether to also return the indices for where elements in the original input ended up in
the returned unique list. Default: ``False`` .
return_counts(bool): Whether to also return the counts for each unique element. Default: ``False`` .
dim(int): the dimension to operate upon. If ``None``, the unique of the flattened input is returned.
Otherwise, each of the tensors indexed by the given dimension is treated as one of the elements to apply the
unique operation upon. Default: ``None`` .
Returns:
A tensor or a tuple of tensors containing some of tensor objects (`output`, `inverse_indices`, `counts`).
- output(Tensor) - The output tensor including the unique elements of input tensor, it has same dtype as input.
- inverse_indices(Tensor) - Return when ``return_inverse`` is True. It represents the indices for where
elements in the original input map to in the output. When ``dim`` is ``None``, it has same shape as input,
otherwise, the shape is input.shape[dim].
- counts(Tensor) - Return when ``return_counts`` is True. It represents the number of occurrences for each
unique value or tensor. When ``dim`` is ``None``, it has same shape as output, otherwise, the shape is
output.shape(dim).
Raises:
TypeError: If `input` is not a Tensor.
Supported Platforms:
``Ascend``
Examples:
>>> import mindspore
>>> import numpy as np
>>> from mindspore import Tensor, nn
>>> from mindspore import ops
>>> x = Tensor(np.array([1, 2, 5, 2]), mindspore.int32)
>>> output = ops.unique_ext(x, return_inverse=True)
>>> print(output)
(Tensor(shape=[3], dtype=Int32, value= [1, 2, 5]), Tensor(shape=[4], dtype=Int64, value= [0, 1, 2, 1]))
>>> y = output[0]
>>> print(y)
[1 2 5]
>>> idx = output[1]
>>> print(idx)
[0 1 2 1]
"""
if not F.isconstant(return_inverse) or not F.isconstant(return_counts):
raise ValueError(
f"For 'unique_ext', 'return_inverse' and 'return_counts' cannot be mutable")
if dim is None:
y, inverse, counts = unique2_(
input, sorted, return_inverse, return_counts)
else:
validator.check_value_type(
"return_counts", return_counts, [bool], "unique_ext")
y, inverse, counts = unique_dim_(input, sorted, return_inverse, dim)
if return_inverse and return_counts:
return y, inverse, counts
if return_inverse:
return y, inverse
if return_counts:
return y, counts
return y
[docs]def unique_with_pad(x, pad_num):
"""
Returns unique elements and relative indexes in 1-D tensor, filled with padding num.
The basic function is the same as the Unique operator, but the UniqueWithPad operator adds a Pad function.
The returned tuple(`y`, `idx`) after the input Tensor `x` is processed by the unique operator,
in which the shapes of `y` and `idx` are mostly not equal. Therefore, in order to solve the above situation,
the UniqueWithPad operator will fill the `y` Tensor with the `pad_num` specified by the user
to make it have the same shape as the Tensor `idx`.
.. warning::
:func:`mindspore.ops.unique_with_pad` is deprecated from version 2.4 and will be removed in a future version.
Please use the :func:`mindspore.ops.unique` combined with :func:`mindspore.ops.pad` to realize
the same function.
Args:
x (Tensor): The tensor need to be unique. Must be 1-D vector with types: int32, int64.
pad_num (int): Pad num. The data type is an int.
Returns:
tuple(Tensor), tuple of 2 tensors, `y` and `idx`.
- y (Tensor) - The unique elements filled with pad_num, the shape and data type same as `x`.
- idx (Tensor) - The index of each value of `x` in the unique output `y`, the shape and data type same as `x`.
Raises:
TypeError: If dtype of `x` is neither int32 nor int64.
ValueError: If `x` is not a 1-D Tensor.
Supported Platforms:
Deprecated
Examples:
>>> import mindspore
>>> import numpy as np
>>> from mindspore import Tensor, nn
>>> from mindspore import ops
>>> x = Tensor(np.array([1, 2, 2, 3, 5, 5]), mindspore.int32)
>>> output = ops.unique_with_pad(x, 0)
>>> print(output)
(Tensor(shape=[6], dtype=Int32, value= [1, 2, 3, 5, 0, 0]),
Tensor(shape=[6], dtype=Int32, value= [0, 1, 1, 2, 3, 3]))
>>> y = output[0]
>>> print(y)
[1 2 3 5 0 0]
>>> idx = output[1]
>>> print(idx)
[0 1 1 2 3 3]
"""
return _get_cache_prim(P.UniqueWithPad)()(x, pad_num)
[docs]def unique_consecutive(input, return_idx=False, return_counts=False, axis=None):
"""
Returns the elements that are unique in each consecutive group of equivalent elements in the input tensor.
Args:
input (Tensor): The input tensor.
return_idx (bool, optional): Whether to return the index of where the element in the original input
maps to the position in the output. Default: ``False`` .
return_counts (bool, optional): Whether to return the counts of each unique element. Default: ``False`` .
axis (int, optional): The dimension to apply unique. If ``None`` , the unique of the flattened input is
returned. If specified, it must be int32 or int64. Default: ``None`` .
Returns:
A tensor or a tuple of tensors containing tensor objects (`output`, `idx`, `counts`). `output` has the
same type as `input` and is used to represent the output list of unique scalar elements. If `return_idx` is
True, there will be an additional returned tensor, `idx`, which has the same shape as `input` and represents
the index of where the element in the original input maps to the position in the output. If `return_counts`
is True, there will be an additional returned tensor, `counts`, which represents the number of occurrences
for each unique value or tensor.
Raises:
TypeError: If `input` is not a Tensor.
TypeError: If dtype of `input` is not supported.
TypeError: If `return_idx` is not a bool.
TypeError: If `return_counts` is not a bool.
TypeError: If `axis` is not an int.
ValueError: If `axis` is not in the range of :math:`[-ndim, ndim-1]`.
Supported Platforms:
``Ascend`` ``GPU`` ``CPU``
Examples:
>>> import numpy as np
>>> from mindspore import Tensor, ops
>>> from mindspore import dtype as mstype
>>> x = Tensor(np.array([1, 1, 2, 2, 3, 1, 1, 2]), mstype.int32)
>>> output, idx, counts = ops.unique_consecutive(x, True, True, None)
>>> print(output)
[1 2 3 1 2]
>>> print(idx)
[0 0 1 1 2 3 3 4]
>>> print(counts)
[2 2 1 2 1]
"""
if not isinstance(input, (Tensor, Tensor_)):
raise TypeError("For 'unique_consecutive', 'input' must be Tensor.")
unique_consecutive_op = _get_cache_prim(
UniqueConsecutive)(return_idx, return_counts, axis)
output, idx, counts = unique_consecutive_op(input)
if return_idx and return_counts:
return output, idx, counts
if return_idx:
return output, idx
if return_counts:
return output, counts
return output
[docs]def searchsorted(sorted_sequence, values, *, out_int32=False, right=False, side=None, sorter=None):
"""
Return the position indices such that after inserting the values into the `sorted_sequence`, the order of innermost
dimension of the `sorted_sequence` remains unchanged.
Args:
sorted_sequence (Tensor): The input tensor.
It must contain a monotonically increasing sequence on the innermost dimension.
values (Tensor): The value that should be inserted.
Keyword Args:
out_int32 (bool, optional): Output datatype. If ``True`` , the output datatype will be int32;
if ``False`` , the output datatype will be int64. Default: ``False`` .
right (bool, optional): Search Strategy. If ``True`` , return the last suitable index found;
if ``False`` , return the first such index. Default: ``False`` .
side (str, optional): the same as right but preferred. ``"left"`` corresponds to ``False`` for `right`
and ``"right"`` corresponds to ``True`` for `right`. An error will be reported if this parameter is
set to ``"left"`` while `right` is ``True``. Default: ``None`` .
sorter(Tensor, optional): if provided, a tensor matching the shape of the unsorted sorted_sequence
containing a sequence of indices that sort it in the ascending order on the innermost
dimension and type must be int64. Default: ``None`` . CPU and GPU can only use default values
Returns:
Tensor containing the indices from the innermost dimension of `sorted_sequence` such that,
if insert the corresponding value in the `values` Tensor, the order of `sorted_sequence` would be preserved,
whose datatype is int32 if out_int32 is ``True`` , otherwise int64, and shape is the same as the shape of
`values`.
Raises:
ValueError: If the dimension of `sorted_sequence` isn't 1 and all dimensions except the last dimension of
`sorted_sequence` and `values` are different.
ValueError: If `sorted_sequence` value is a scalar.
ValueError: If `values` is a scalar when `sorted_sequence` dimension is not 1.
Supported Platforms:
``Ascend`` ``GPU`` ``CPU``
Examples:
>>> import mindspore
>>> import numpy as np
>>> from mindspore import Tensor, ops
>>> sorted_sequence = Tensor(np.array([[0, 1, 3, 5, 7], [2, 4, 6, 8, 10]]), mindspore.float32)
>>> values = Tensor(np.array([[3, 6, 9], [3, 6, 9]]), mindspore.float32)
>>> output = ops.searchsorted(sorted_sequence, values)
>>> print(output)
[[2 4 5]
[1 2 4]]
"""
validator.check_value_type("out_int32", out_int32, [bool], "search_sorted")
validator.check_value_type("right", right, [bool], "search_sorted")
dtype = mstype.int32 if bool(out_int32) else mstype.int64
if (side == "left" and right is True):
raise ValueError(f"For 'searchsorted', side and right can't be set to opposites,"
f"got side of left while right was True.")
if side == "right":
right = True
return search_sorted_(sorted_sequence, values, sorter, dtype, right)
[docs]def ger(input, vec2):
r"""
Ger product of `input` and `vec2`. Calculate the outer product of two arrays. If `input` is a 1D Tensor of
shape :math:`(m,)` and `vec2` is a 1D Tensor of shape :math:`(n,)`, then `output` must be a 2D Tensor of shape
:math:`(m, n)`.
Note:
Currently Ascend does not support float64 data input.
Args:
input (Tensor): input Tensor, with dtype of float16, float32 or float64.
vec2 (Tensor): input Tensor, with dtype of float16, float32 or float64, must have the same dtype as `input`.
Returns:
Tensor, output matrix with the same dtype as inputs. With `input` shape :math:`(m,)` and
`vec2` shape of :math:`(n,)`, the `output` has shape :math:`(m, n)`.
Raises:
TypeError: If `input` or `vec2` is not a 1-D Tensor.
TypeError: If the dtype of `input` and `vec2` is not float16, float32 or float64.
TypeError: If the dtype of `input` and `vec2` are not the same.
Supported Platforms:
``Ascend`` ``GPU`` ``CPU``
Examples:
>>> import mindspore
>>> from mindspore import Tensor, ops
>>> input = Tensor([1., 2., 3., 4.], mindspore.float32)
>>> vec2 = Tensor([1., 2., 3.], mindspore.float32)
>>> output = ops.ger(input, vec2)
>>> print(output)
[[ 1. 2. 3.]
[ 2. 4. 6.]
[ 3. 6. 9.]
[ 4. 8. 12.]]
"""
return ger_(input, vec2)
[docs]def size(input_x):
r"""
Returns a Scalar of type int that represents the size of the input Tensor and the total number of elements in the
Tensor.
Args:
input_x (Tensor): Input parameters, the shape of tensor is :math:`(x_1, x_2, ..., x_R)`. The data type is
`number <https://www.mindspore.cn/docs/en/r2.4.1/api_python/mindspore/mindspore.dtype.html>`_.
Returns:
int. A scalar representing the elements' size of `input_x`, tensor is the number of elements
in a tensor, :math:`size=x_1*x_2*...x_R`. The data type is an int.
Raises:
TypeError: If `input_x` is not a Tensor.
Supported Platforms:
``Ascend`` ``GPU`` ``CPU``
Examples:
>>> import mindspore
>>> import numpy as np
>>> from mindspore import Tensor, ops
>>> input_x = Tensor(np.array([[2, 2], [2, 2]]), mindspore.float32)
>>> output = ops.size(input_x)
>>> print(output)
4
"""
return size_(input_x)
[docs]def shape(input_x):
"""
Returns the shape of the input tensor.
Args:
input_x (Tensor): The shape of tensor is :math:`(x_1, x_2, ..., x_R)`.
Returns:
tuple[int], the output tuple is constructed by multiple integers,
:math:`(x_1, x_2, ..., x_R)`.
Raises:
TypeError: If `input_x` is not a Tensor.
Supported Platforms:
``Ascend`` ``GPU`` ``CPU``
Examples:
>>> import mindspore
>>> import numpy as np
>>> from mindspore import Tensor, ops
>>> input_x = Tensor(np.ones(shape=[3, 2, 1]), mindspore.float32)
>>> output = ops.shape(input_x)
>>> print(output)
(3, 2, 1)
"""
return shape_(input_x)
[docs]def dyn_shape(input_x):
"""
Returns the shape of the input tensor.
Args:
input_x (Tensor): The input Tensor.
Returns:
Tensor, the shape of `input_x` .
Raises:
TypeError: If `input_x` is not a Tensor.
Supported Platforms:
``Ascend`` ``GPU`` ``CPU``
Examples:
>>> import mindspore
>>> import numpy as np
>>> from mindspore import Tensor, ops
>>> input_x = Tensor(np.ones(shape=[3, 2, 1]), mindspore.float32)
>>> output = ops.dyn_shape(input_x)
>>> print(output)
[3 2 1]
"""
return tensor_shape_(input_x)
[docs]def reverse_sequence(x, seq_lengths, seq_dim, batch_dim=0):
r"""
Reverses variable length slices.
Args:
x (Tensor): The input to reverse, supporting all number types including bool.
seq_lengths (Tensor): Specified reversing length, must be a 1-D vector with int32 or int64 types.
seq_dim (int): The dimension where reversal is performed. Required.
batch_dim (int): The input is sliced in this dimension. Default: ``0`` .
Returns:
Tensor, with the same shape and data type as `x`.
Raises:
TypeError: If `seq_dim` or `batch_dim` is not an int.
ValueError: If :math:`len(seq\_lengths) != x.shape[batch\_dim]`.
ValueError: If :math:`batch\_dim == seq\_dim`.
ValueError: If :math:`seq\_dim < 0` or :math:`seq\_dim >= len(x.shape)`.
ValueError: If :math:`batch\_dim < 0` or :math:`batch\_dim >= len(x.shape)`.
RuntimeError: If any value of `seq_lengths` is less than 0.
RuntimeError: If any value of `seq_lengths` is larger than `x.shape[seq_dim]`.
Supported Platforms:
``Ascend`` ``GPU`` ``CPU``
Examples:
>>> import mindspore
>>> import numpy as np
>>> from mindspore import Tensor, ops
>>> x = Tensor(np.array([[1, 2, 3], [4, 5, 6], [7, 8, 9]]), mindspore.float32)
>>> seq_lengths = Tensor(np.array([1, 2, 3]))
>>> output = ops.reverse_sequence(x, seq_lengths, seq_dim=1)
>>> print(output)
[[1. 2. 3.]
[5. 4. 6.]
[9. 8. 7.]]
>>> x = Tensor(np.array([[1, 2, 3], [4, 5, 6], [7, 8, 9]]), mindspore.float32)
>>> seq_lengths = Tensor(np.array([1, 2, 3]))
>>> output = ops.reverse_sequence(x, seq_lengths, seq_dim=0, batch_dim=1)
>>> print(output)
[[1. 5. 9.]
[4. 2. 6.]
[7. 8. 3.]]
>>> x = Tensor(np.array([[1, 2, 3], [4, 5, 6], [7, 8, 9]]), mindspore.float32)
>>> seq_lengths = Tensor(np.array([2, 2, 3]))
>>> output = ops.reverse_sequence(x, seq_lengths, seq_dim=1)
>>> print(output)
[[2. 1. 3.]
[5. 4. 6.]
[9. 8. 7.]]
>>> x = Tensor(np.array([[1, 2, 3], [4, 5, 6], [7, 8, 9]]), mindspore.float32)
>>> seq_lengths = Tensor(np.array([3, 2, 3]))
>>> output = ops.reverse_sequence(x, seq_lengths, seq_dim=1)
>>> print(output)
[[3. 2. 1.]
[5. 4. 6.]
[9. 8. 7.]]
>>> x = Tensor(np.array([[1, 2, 3, 4], [5, 6, 7, 8]]), mindspore.float32)
>>> seq_lengths = Tensor(np.array([4, 4]))
>>> output = ops.reverse_sequence(x, seq_lengths, seq_dim=1)
>>> print(output)
[[4. 3. 2. 1.]
[8. 7. 6. 5.]]
"""
return _get_cache_prim(P.ReverseSequence)(seq_dim=seq_dim, batch_dim=batch_dim)(x, seq_lengths)
[docs]def flatten(input, order='C', *, start_dim=1, end_dim=-1):
r"""
Flatten a tensor along dimensions from `start_dim` to `start_dim`.
Args:
input (Tensor): The input Tensor.
order (str, optional): Only ``'C'`` and ``'F'`` are supported.
``'C'`` means to flatten in row-major (C-style) order.
``'F'`` means to flatten in column-major (Fortran-style) order. Default: ``'C'`` .
Keyword Args:
start_dim (int, optional): The first dimension to flatten. Default: ``1`` .
end_dim (int, optional): The last dimension to flatten. Default: ``-1`` .
Returns:
Tensor. If no dimensions are flattened, returns the original `input`, otherwise return the flattened Tensor.
If `input` is a 0-dimensional Tensor, a 1-dimensional Tensor will be returned.
Raises:
TypeError: If `input` is not a Tensor.
TypeError: If `order` is not string type.
ValueError: If `order` is string type, but not ``'C'`` or ``'F'``.
TypeError: If `start_dim` or `end_dim` is not int.
ValueError: If `start_dim` is greater than `end_dim` after canonicalized.
ValueError: If `start_dim` or `end_dim` is not in range of [-input.dim, input.dim-1].
Supported Platforms:
``Ascend`` ``GPU`` ``CPU``
Examples:
>>> import mindspore
>>> import numpy as np
>>> from mindspore import Tensor, ops
>>> input_x = Tensor(np.ones(shape=[1, 2, 3, 4]), mindspore.float32)
>>> output = ops.flatten(input_x)
>>> print(output.shape)
(1, 24)
"""
def check_axis_valid(axis, ndim):
if axis < -ndim or axis >= ndim:
raise ValueError("'start_dim' or 'end_dim' out of range.")
def check_dim_valid(start_dim, end_dim):
if start_dim > end_dim:
raise ValueError(
"For 'flatten', 'start_dim' cannot come after 'end_dim'.")
def canonicalize_axis(axis, x_rank):
ndim = x_rank if x_rank != 0 else 1
check_axis_valid(axis, ndim)
return axis if axis >= 0 else axis + ndim
# Check the types of arguments.
if not isinstance(input, Tensor):
raise TypeError(f"For 'flatten', argument 'input' must be Tensor.")
if not isinstance(start_dim, int) or not isinstance(end_dim, int) or \
isinstance(start_dim, bool) or isinstance(end_dim, bool):
raise TypeError(
f"For 'flatten', both 'start_dim' and 'end_dim' must be int.")
check_flatten_order_const(order)
if order == 'F':
x_rank = rank_(input)
# If input is a 0-dimensional Tensor, a 1-dimensional Tensor will be returned.
if x_rank in (0, 1):
return reshape_(input, (-1,))
perm = ops.make_range(0, x_rank)
new_order = ops.tuple_reversed(perm)
input = transpose_(input, new_order)
# Handle the default case.
x_shape = shape_(input)
x_rank = rank_(input)
if start_dim == 1 and end_dim == -1:
if x_rank in (0, 1):
return reshape_(input, (-1,))
return flatten_(input)
# Check axis.
start_dim = canonicalize_axis(start_dim, x_rank)
end_dim = canonicalize_axis(end_dim, x_rank)
check_dim_valid(start_dim, end_dim)
# If input is a 0-dimensional Tensor, a 1-dimensional Tensor will be returned.
if x_rank in (0, 1):
return reshape_(input, (-1,))
# If no dimensions to flatten, return the original object.
if start_dim == end_dim:
return input
# Flatten elements along specified dimensions.
dim_length = 1
idx = start_dim
while idx <= end_dim:
dim_length *= x_shape[idx]
idx += 1
new_shape = x_shape[:start_dim] + (dim_length,) + x_shape[end_dim + 1:]
return reshape_(input, new_shape)
[docs]def slice(input_x, begin, size):
r"""
Slices a tensor in the specified shape.
Slice the tensor `input_x` in shape of `size` and starting at the location specified by `begin`.
The slice `begin` represents the offset in each dimension of `input_x`.
The slice `size` represents the size of the output tensor.
Note:
`begin` is zero-based and `size` is one-based.
If `size[i]` is -1, all remaining elements in dimension i are included in the slice.
This is equivalent to setting :math:`size[i] = input\_x.shape(i) - begin[i]`.
Args:
input_x (Tensor): The target tensor.
begin (Union[tuple, list]): The beginning of the slice. Only constant value(>=0) is allowed.
size (Union[tuple, list]): The size of the slice. Only constant value is allowed.
Returns:
Tensor, the shape is input `size`, the data type is the same as `input_x`.
Raises:
TypeError: If `begin` or `size` is neither tuple nor list.
Supported Platforms:
``Ascend`` ``GPU`` ``CPU``
Examples:
>>> from mindspore import Tensor
>>> from mindspore import ops
>>> import numpy as np
>>> data = Tensor(np.array([[[1, 1, 1], [2, 2, 2]],
... [[3, 3, 3], [4, 4, 4]],
... [[5, 5, 5], [6, 6, 6]]]).astype(np.int32))
>>> output = ops.slice(data, (1, 0, 0), (1, 1, 3))
>>> print(output)
[[[3 3 3]]]
>>> output = ops.slice(data, (1, 0, 0), (1, 1, 2))
>>> print(output)
[[[3 3]]]
>>> output = ops.slice(data, (1, 0, 0), (1, 1, 1))
>>> print(output)
[[[3]]]
>>> output = ops.slice(data, (1, 1, 0), (1, 1, 3))
>>> print(output)
[[[4 4 4]]]
>>> output = ops.slice(data, (1, 0, 1), (1, 1, 2))
>>> print(output)
[[[3 3]]]
"""
return tensor_slice(input_x, begin, size)
[docs]def stack(tensors, axis=0):
r"""
Stacks a list of tensors in specified axis.
Stacks the list of input tensors with the same rank `R`, output is a tensor of rank `(R+1)`.
Given input tensors of shape :math:`(x_1, x_2, ..., x_R)`. Set the number of input tensors as `N`.
If :math:`axis \ge 0`, the shape of the output tensor is
:math:`(x_1, x_2, ..., x_{axis}, N, x_{axis+1}, ..., x_R)`.
Args:
tensors (Union[tuple, list]): A Tuple or list of Tensor objects with the same shape and type.
axis (int): Dimension to stack. The range is [-(R+1), R+1). Default: ``0`` .
Returns:
Tensor. A stacked Tensor with the same type as `tensors`.
Raises:
TypeError: If the data types of elements in `tensors` are not the same.
ValueError: If the length of `tensors` is not greater than zero;
or if axis is out of the range [-(R+1), R+1);
or if the shapes of elements in tensors are not the same.
Supported Platforms:
``Ascend`` ``GPU`` ``CPU``
Examples:
>>> import numpy as np
>>> from mindspore import Tensor, ops
>>> input_x1 = Tensor(np.array([0, 1]).astype(np.float32))
>>> input_x2 = Tensor(np.array([2, 3]).astype(np.float32))
>>> output = ops.stack((input_x1, input_x2), 0)
>>> print(output)
[[0. 1.]
[2. 3.]]
"""
_stack = _get_cache_prim(P.Stack)(axis)
return _stack(tensors)
[docs]def unstack(input_x, axis=0):
r"""
Unstacks tensor in specified axis, this is the opposite of :func:`mindspore.ops.stack`.
Assuming input is a tensor of rank `R`, output tensors will have rank `(R-1)`.
Args:
input_x (Tensor): The shape is :math:`(x_1, x_2, ..., x_R)`.
A tensor to be unstacked and the rank of the tensor must be greater than 0.
axis (int): Dimension along which to unpack. Default: ``0`` .
Negative values wrap around. The range is [-R, R).
Returns:
A tuple of tensors, the shape of each objects is the same.
Given a tensor of shape :math:`(x_1, x_2, ..., x_R)`. If :math:`0 \le axis`,
the shape of tensor in output is :math:`(x_1, x_2, ..., x_{axis}, x_{axis+2}, ..., x_R)`.
Raises:
ValueError: If axis is out of the range [-len(input_x.shape), len(input_x.shape)).
Supported Platforms:
``Ascend`` ``GPU`` ``CPU``
Examples:
>>> import numpy as np
>>> from mindspore import Tensor, ops
>>> input_x = Tensor(np.array([[1, 1, 1, 1], [2, 2, 2, 2]]))
>>> output = ops.unstack(input_x, 0)
>>> print(output)
(Tensor(shape=[4], dtype=Int64, value= [1, 1, 1, 1]), Tensor(shape=[4], dtype=Int64, value= [2, 2, 2, 2]))
"""
_unstack = _get_cache_prim(P.Unstack)(axis)
return _unstack(input_x)
[docs]def unbind(input, dim=0):
r"""
Removes a tensor dimension in specified axis.
Unstacks a tensor of rank `R` along axis dimension, and output tensors will have rank `(R-1)`.
Given a tensor of shape :math:`(n_1, n_2, ..., n_R)` and a specified `dim`,
shape of the output tensors is :math:`(n_1, n_2, ..., n_{dim}, n_{dim+2}, ..., n_R)`.
Args:
input (Tensor): The shape is :math:`(n_1, n_2, ..., n_R)`.
A tensor to be unstacked and the rank of the tensor must be greater than 0.
dim (int): Dimension along which to unpack. Negative values wrap around. The range is [-R, R). Default: ``0`` .
Returns:
A tuple of tensors, the shape of each objects is the same.
Raises:
ValueError: If axis is out of the range [-R, R).
Supported Platforms:
``Ascend`` ``GPU`` ``CPU``
Examples:
>>> import numpy as np
>>> from mindspore import Tensor, ops
>>> x = Tensor(np.array([[1, 2, 3], [4, 5, 6], [7, 8, 9]]))
>>> output = ops.unbind(x, dim=0)
>>> print(output)
(Tensor(shape=[3], dtype=Int64, value=[1, 2, 3]), Tensor(shape=[3], dtype=Int64, value=[4, 5, 6]),
Tensor(shape=[3], dtype=Int64, value=[7, 8, 9]))
"""
_unstack = _get_cache_prim(P.Unstack)(dim)
return _unstack(input)
[docs]def unsqueeze(input, dim):
"""
Adds an additional dimension to `input` at the given dim.
Args:
input (Tensor): The shape of tensor is :math:`(n_1, n_2, ..., n_R)`.
dim (int): Specifies the dimension index at which to expand
the shape of `input`. The value of `dim` must be in the range
`[-input.ndim-1, input.ndim]`. Only constant value is allowed.
Returns:
Tensor, the shape of tensor is :math:`(1, n_1, n_2, ..., n_R)` if the
value of `dim` is 0. It has the same data type as `input`.
Raises:
TypeError: If `dim` is not an int.
ValueError: If `dim` is not in the valid range :math:`[-input.ndim-1, input.ndim]`.
Supported Platforms:
``Ascend`` ``GPU`` ``CPU``
Examples:
>>> import mindspore
>>> import numpy as np
>>> from mindspore import Tensor, ops
>>> input_tensor = Tensor(np.array([[2, 2], [2, 2]]), mindspore.float32)
>>> output = ops.unsqueeze(input_tensor, dim=0)
>>> print(output)
[[[2. 2.]
[2. 2.]]]
"""
return expand_dims(input, dim)
[docs]def squeeze(input, axis=None):
"""
Return the Tensor after deleting the dimension of size 1 in the specified `axis`.
If :math:`axis=None`, it will remove all the dimensions of size 1.
If `axis` is specified, it will remove the dimensions of size 1 in the given `axis`.
For example, if the dimension is not specified :math:`axis=None`, input shape is (A, 1, B, C, 1, D),
then the shape of the output Tensor is (A, B, C, D). If the dimension is specified, the squeeze operation
is only performed in the specified dimension. If input shape is (A, 1, B), input Tensor will be changed
to (A, B) when :math:`axis=1`, but when :math:`axis=0` or :math:`axis=2`, an error will occur.
Note:
- Squeezing a dimension that is not 1 will raise an error.
- Please note that in dynamic graph mode, the output Tensor will share data with the input Tensor,
and there is no Tensor data copy process.
- The dimension index starts at 0 and must be in the range `[-input.ndim, input.ndim]`.
Args:
input (Tensor): The shape of tensor is :math:`(x_1, x_2, ..., x_R)`.
axis (Union[int, tuple(int), list(int)]): Specifies the dimension indexes of shape to be removed, which will
remove all the dimensions of size 1 in the given axis parameter. If specified, it must be int32 or int64.
Default: ``None`` , an empty tuple will be used.
Returns:
Tensor, the shape of tensor is :math:`(x_1, x_2, ..., x_S)`.
Raises:
TypeError: If `input` is not a tensor.
TypeError: If `axis` is not an int, tuple or list.
TypeError: If `axis` is a tuple or list whose elements are not all int.
ValueError: If the corresponding dimension of the specified axis isn't equal to 1.
Supported Platforms:
``Ascend`` ``GPU`` ``CPU``
Examples:
>>> import mindspore
>>> import numpy as np
>>> from mindspore import Tensor, ops
>>> input = Tensor(np.ones(shape=[3, 2, 1]), mindspore.float32)
>>> output = ops.squeeze(input)
>>> print(output)
[[1. 1.]
[1. 1.]
[1. 1.]]
"""
if axis is None:
axis = ()
if isinstance(axis, list):
axis = tuple(axis)
squeeze_ = _get_cache_prim(P.Squeeze)(axis)
return squeeze_(input)
[docs]def scatter_mul(input_x, indices, updates):
r"""
Using given values to update tensor value through the mul operation, along with the input indices.
This operation outputs the `input_x` after the update is done, which makes it convenient to use the updated value.
for each `i, ..., j` in `indices.shape`:
.. math::
\text{input_x}[\text{indices}[i, ..., j], :] \mathrel{*}= \text{updates}[i, ..., j, :]
Inputs of `input_x` and `updates` comply with the implicit type conversion rules to make the data types consistent.
If they have different data types, the lower priority data type will be converted to
the relatively highest priority data type. A RuntimeError will be reported
when the data types of parameters need to be converted.
Args:
input_x (Parameter): The target tensor to be updated, with data type of Parameter.
The shape is :math:`(N,*)` where :math:`*` means any number of additional dimensions.
indices (Tensor): The index to do mul operation whose data type must be int32 or int64.
updates (Tensor): The tensor doing the mul operation with `input_x`,
the data type is same as `input_x`, the shape is `indices.shape + input_x.shape[1:]`.
Returns:
Tensor, the updated `input_x`, has the same shape and type as `input_x`.
Raises:
TypeError: If `indices` is not an int32 or int64.
ValueError: If the shape of `updates` is not equal to `indices.shape + input_x.shape[1:]`.
RuntimeError: If the data type of `input_x` and `updates` conversion of Parameter
is required when data type conversion of Parameter is not supported.
Supported Platforms:
``Ascend`` ``GPU`` ``CPU``
Examples:
>>> import mindspore
>>> import numpy as np
>>> from mindspore import Tensor, ops, Parameter
>>> input_x = Parameter(Tensor(np.array([[1.0, 1.0, 1.0], [2.0, 2.0, 2.0]]), mindspore.float32), name="x")
>>> indices = Tensor(np.array([0, 1]), mindspore.int32)
>>> updates = Tensor(np.array([[2.0, 2.0, 2.0], [2.0, 2.0, 2.0]]), mindspore.float32)
>>> output = ops.scatter_mul(input_x, indices, updates)
>>> print(output)
[[2. 2. 2.]
[4. 4. 4.]]
>>> # for input_x will be updated after the operation is completed. input_x need to be re-initialized.
>>> input_x = Parameter(Tensor(np.array([[1.0, 1.0, 1.0], [2.0, 2.0, 2.0]]), mindspore.float32), name="x")
>>> # for indices = [[0, 1], [1, 1]]
>>> # step 1: [0, 1]
>>> # input_x[0] = [1.0, 1.0, 1.0] * [1.0, 1.0, 1.0] = [1.0, 1.0, 1.0]
>>> # input_x[1] = [2.0, 2.0, 2.0] * [3.0, 3.0, 3.0] = [6.0, 6.0, 6.0]
>>> # step 2: [1, 1]
>>> # input_x[1] = [6.0, 6.0, 6.0] * [7.0, 7.0, 7.0] = [42.0, 42.0, 42.0]
>>> # input_x[1] = [42.0, 42.0, 42.0] * [9.0, 9.0, 9.0] = [378.0, 378.0, 378.0]
>>> indices = Tensor(np.array([[0, 1], [1, 1]]), mindspore.int32)
>>> updates = Tensor(np.array([[[1.0, 1.0, 1.0], [3.0, 3.0, 3.0]],
... [[7.0, 7.0, 7.0], [9.0, 9.0, 9.0]]]), mindspore.float32)
>>> output = ops.scatter_mul(input_x, indices, updates)
>>> print(output)
[[ 1. 1. 1.]
[378. 378. 378.]]
>>> # for input_x will be updated after the operation is completed. input_x need to be re-initialized.
>>> input_x = Parameter(Tensor(np.array([[1.0, 1.0, 1.0], [2.0, 2.0, 2.0]]), mindspore.float32), name="x")
>>> # for indices = [[1, 0], [1, 1]]
>>> # step 1: [1, 0]
>>> # input_x[0] = [1.0, 1.0, 1.0] * [3.0, 3.0, 3.0] = [3.0, 3.0, 3.0]
>>> # input_x[1] = [2.0, 2.0, 2.0] * [1.0, 1.0, 1.0] = [2.0, 2.0, 2.0]
>>> # step 2: [1, 1]
>>> # input_x[1] = [2.0, 2.0, 2.0] * [7.0, 7.0, 7.0] = [14.0, 14.0, 14.0]
>>> # input_x[1] = [14.0, 14.0, 14.0] * [9.0, 9.0, 9.0] = [126.0, 126.0, 126.0]
>>> indices = Tensor(np.array([[1, 0], [1, 1]]), mindspore.int32)
>>> updates = Tensor(np.array([[[1.0, 1.0, 1.0], [3.0, 3.0, 3.0]],
... [[7.0, 7.0, 7.0], [9.0, 9.0, 9.0]]]), mindspore.float32)
>>> output = ops.scatter_mul(input_x, indices, updates)
>>> print(output)
[[ 3. 3. 3.]
[126. 126. 126.]]
>>> # for input_x will be updated after the operation is completed. input_x need to be re-initialized.
>>> input_x = Parameter(Tensor(np.array([[1.0, 1.0, 1.0], [2.0, 2.0, 2.0]]), mindspore.float32), name="x")
>>> # for indices = [[0, 1], [0, 1]]
>>> # step 1: [0, 1]
>>> # input_x[0] = [1.0, 1.0, 1.0] * [1.0, 1.0, 1.0] = [1.0, 1.0, 1.0]
>>> # input_x[1] = [2.0, 2.0, 2.0] * [3.0, 3.0, 3.0] = [6.0, 6.0, 6.0]
>>> # step 2: [0, 1]
>>> # input_x[0] = [1.0, 1.0, 1.0] * [7.0, 7.0, 7.0] = [7.0, 7.0, 7.0]
>>> # input_x[1] = [6.0, 6.0, 6.0] * [9.0, 9.0, 9.0] = [54.0, 54.0, 54.0]
>>> indices = Tensor(np.array([[0, 1], [0, 1]]), mindspore.int32)
>>> updates = Tensor(np.array([[[1.0, 1.0, 1.0], [3.0, 3.0, 3.0]],
... [[7.0, 7.0, 7.0], [9.0, 9.0, 9.0]]]), mindspore.float32)
>>> output = ops.scatter_mul(input_x, indices, updates)
>>> print(output)
[[ 7. 7. 7.]
[54. 54. 54.]]
"""
return scatter_mul_(input_x, indices, updates)
[docs]def scatter_max(input_x, indices, updates):
r"""
Using given values to update tensor value through the max operation, along with the input indices.
This operation outputs the `input_x` after the update is done, which makes it convenient to use the updated value.
for each :math:`i, ..., j` in `indices.shape`:
.. math::
\text{input_x}[\text{indices}[i, ..., j], :]
= \max(\text{input_x}[\text{indices}[i, ..., j], :], \text{updates}[i, ..., j, :])
Inputs of `input_x` and `updates` follow the implicit type conversion rules to keep the data types consistent.
If they have different data types, the lower priority data type will be converted to the relatively highest
priority data type. A RuntimeError will be reported when `updates` does not support conversion to the data type
required by `input_x`.
Args:
input_x (Parameter): The target tensor, with data type of Parameter.
The shape is :math:`(N, *)` where :math:`*` means, any number of additional dimensions.
indices (Tensor): The index to do max operation whose data type must be mindspore.int32.
updates (Tensor): The tensor doing the max operation with `input_x`,
the data type is same as `input_x`, the shape is `indices.shape + x.shape[1:]`.
Returns:
Tensor, the updated `input_x`, the type and shape same as `input_x`.
Raises:
TypeError: If `indices` is not an int32 or int64.
ValueError: If the shape of `updates` is not equal to `indices.shape + input_x.shape[1:]`.
RuntimeError: If the data type of `input_x` and `updates` conversion of Parameter
is required when data type conversion of Parameter is not supported.
RuntimeError: On the Ascend platform, the input data dimension of `input_x` , `indices`
and `updates` is greater than 8 dimensions.
Supported Platforms:
``Ascend`` ``GPU`` ``CPU``
Examples:
>>> import mindspore
>>> import numpy as np
>>> from mindspore import Tensor, ops, Parameter
>>> input_x = Parameter(Tensor(np.array([[1.0, 2.0, 3.0], [4.0, 5.0, 6.0]]), mindspore.float32), name="input_x")
>>> indices = Tensor(np.array([[0, 0], [1, 1]]), mindspore.int32)
>>> updates = Tensor(np.ones([2, 2, 3]) * 88, mindspore.float32)
>>> output = ops.scatter_max(input_x, indices, updates)
>>> print(output)
[[88. 88. 88.]
[88. 88. 88.]]
"""
return scatter_max_(input_x, indices, updates)
[docs]def scatter_add(input_x, indices, updates):
r"""
Using given values to update tensor value through the add operation, along with the input indices.
This operation outputs the `input_x` after the update is done, which makes it convenient to use the updated value.
Args:
input_x (Parameter): The target tensor, with data type of Parameter.
indices (Tensor): The index to do add operation whose data type must be int32 or int64.
updates (Tensor): The tensor doing the add operation with `input_x`,
the data type is same as `input_x`, the shape is `indices.shape + x.shape[1:]`.
Returns:
Tensor, the updated `input_x`, has the same shape and type as `input_x`.
Raises:
TypeError: If `indices` is not an int32 or int64.
ValueError: If the shape of `updates` is not equal to `indices.shape + input_x.shape[1:]`.
RuntimeError: If the data type of `input_x` and `updates` conversion of Parameter
is required when data type conversion of Parameter is not supported.
Supported Platforms:
``Ascend`` ``GPU`` ``CPU``
Examples:
>>> import numpy as np
>>> import mindspore
>>> from mindspore import Tensor, Parameter
>>> from mindspore import ops
>>> input_x = Parameter(Tensor(np.array([[0.0, 0.0, 0.0], [0.0, 0.0, 0.0]]), mindspore.float32), name="x")
>>> indices = Tensor(np.array([[0, 1], [1, 1]]), mindspore.int32)
>>> updates = Tensor(np.array([[[1.0, 1.0, 1.0], [3.0, 3.0, 3.0]],
... [[7.0, 7.0, 7.0], [9.0, 9.0, 9.0]]]), mindspore.float32)
>>> output = ops.scatter_add(input_x, indices, updates)
>>> print(output)
[[ 1. 1. 1.]
[19. 19. 19.]]
"""
return scatter_add_(input_x, indices, updates)
[docs]def scatter_min(input_x, indices, updates):
r"""
Using given values to update tensor value through the min operation, along with the input indices.
This operation outputs the `input_x` after the update is done, which makes it convenient to use the updated value.
for each :math:`i, ..., j` in `indices.shape`:
.. math::
\text{input_x}[\text{indices}[i, ..., j], :]
= \min(\text{input_x}[\text{indices}[i, ..., j], :], \text{updates}[i, ..., j, :])
Inputs of `input_x` and `updates` comply with the implicit type conversion rules to make the data types consistent.
If they have different data types, the lower priority data type will be converted to
the relatively highest priority data type. A RuntimeError will be reported
when `updates` does not support conversion to the data type required by `input_x`.
Args:
input_x (Parameter): The target tensor, with data type of Parameter.
indices (Tensor): The index to do min operation whose data type must be mindspore.int32 or mindspore.int64.
updates (Tensor): The tensor doing the min operation with `input_x`,
the data type is same as `input_x`, the shape is `indices.shape + input_x.shape[1:]`.
Returns:
Tensor, the updated `input_x`, has the same shape and type as `input_x`.
Raises:
TypeError: If `indices` is not an int32 or an int64.
ValueError: If the shape of `updates` is not equal to `indices.shape + input_x.shape[1:]`.
RuntimeError: If the data type of `input_x` and `updates` conversion of Parameter
is required when data type conversion of Parameter is not supported.
RuntimeError: On the Ascend platform, the input data dimension of `input_x` , `indices`
and `updates` is greater than 8 dimensions.
Supported Platforms:
``Ascend`` ``GPU`` ``CPU``
Examples:
>>> import numpy as np
>>> import mindspore
>>> from mindspore import Tensor, Parameter
>>> from mindspore import ops
>>> input_x = Parameter(Tensor(np.zeros((2, 3)), mindspore.float32), name="input_x")
>>> indices = Tensor(np.array([1, 0]), mindspore.int32)
>>> update = Tensor(np.arange(6).reshape((2, 3)), mindspore.float32)
>>> output = ops.scatter_min(input_x, indices, update)
>>> print(output)
[[0. 0. 0.]
[0. 0. 0.]]
"""
return scatter_min_(input_x, indices, updates)
[docs]def scatter_div(input_x, indices, updates):
r"""
Using given values to update tensor value through the div operation, along with the input indices.
This operation outputs the `input_x` after the update is done, which makes it convenient to use the updated value.
for each :math:`i, ..., j` in `indices.shape`:
.. math::
\text{input_x}[\text{indices}[i, ..., j], :] \mathrel{/}= \text{updates}[i, ..., j, :]
Inputs of `input_x` and `updates` comply with the implicit type conversion rules to make the data types consistent.
If they have different data types, the lower priority data type will be converted to
the relatively highest priority data type. A RuntimeError will be reported
when `updates` does not support conversion to the data type required by `input_x`.
Args:
input_x (Parameter): The target tensor, with data type of Parameter.
indices (Tensor): The index to do divide operation whose data type must be mindspore.int32 or
mindspore.int64.
updates (Tensor): The tensor doing the divide operation with `input_x`, the data type is same as `input_x`,
the shape is `indices.shape + input_x.shape[1:]`.
Returns:
Tensor, the updated `input_x`, has the same shape and type as `input_x`.
Raises:
TypeError: If the type of `indices` is not one of the following dtype: int32, int64.
ValueError: If the shape of `updates` is not equal to `indices.shape + input_x.shape[1:]`.
RuntimeError: If the data type of `input_x` and `updates` conversion of Parameter is required
when data type conversion of Parameter is not supported.
RuntimeError: On the Ascend platform, the input data dimension of `input_x` , `indices`
and `updates` is greater than 8 dimensions.
Supported Platforms:
``Ascend`` ``GPU`` ``CPU``
Examples:
>>> import mindspore
>>> import numpy as np
>>> from mindspore import Tensor, ops, Parameter
>>> input_x = Parameter(Tensor(np.array([[6.0, 6.0, 6.0], [2.0, 2.0, 2.0]]), mindspore.float32), name="x")
>>> indices = Tensor(np.array([0, 1]), mindspore.int32)
>>> updates = Tensor(np.array([[2.0, 2.0, 2.0], [2.0, 2.0, 2.0]]), mindspore.float32)
>>> output = ops.scatter_div(input_x, indices, updates)
>>> print(output)
[[3. 3. 3.]
[1. 1. 1.]]
>>> # for input_x will be updated after the operation is completed. input_x need to be re-initialized.
>>> input_x = Parameter(Tensor(np.array([[105.0, 105.0, 105.0],
... [315.0, 315.0, 315.0]]), mindspore.float32), name="x")
>>> # for indices = [[0, 1], [1, 1]]
>>> # step 1: [0, 1]
>>> # input_x[0] = [105.0, 105.0, 105.0] / [1.0, 1.0, 1.0] = [105.0, 105.0, 105.0]
>>> # input_x[1] = [315.0, 315.0, 315.0] / [3.0, 3.0, 3.0] = [105.0, 105.0, 105.0]
>>> # step 2: [1, 1]
>>> # input_x[1] = [105.0, 105.0, 105.0] / [5.0, 5.0, 5.0] = [21.0, 21.0, 21.0]
>>> # input_x[1] = [21.0, 21.0, 21.0] / [7.0, 7.0, 7.0] = [3.0, 3.0, 3.0]
>>> indices = Tensor(np.array([[0, 1], [1, 1]]), mindspore.int32)
>>> updates = Tensor(np.array([[[1.0, 1.0, 1.0], [3.0, 3.0, 3.0]],
... [[5.0, 5.0, 5.0], [7.0, 7.0, 7.0]]]), mindspore.float32)
>>> output = ops.scatter_div(input_x, indices, updates)
>>> print(output)
[[105. 105. 105.]
[ 3. 3. 3.]]
>>> # for input_x will be updated after the operation is completed. input_x need to be re-initialized.
>>> input_x = Parameter(Tensor(np.array([[105.0, 105.0, 105.0],
... [315.0, 315.0, 315.0]]), mindspore.float32), name="x")
>>> # for indices = [[1, 0], [1, 1]]
>>> # step 1: [1, 0]
>>> # input_x[0] = [105.0, 105.0, 105.0] / [3.0, 3.0, 3.0] = [35.0, 35.0, 35.0]
>>> # input_x[1] = [315.0, 315.0, 315.0] / [1.0, 1.0, 1.0] = [315.0, 315.0, 315.0]
>>> # step 2: [1, 1]
>>> # input_x[1] = [315.0, 315.0, 315.0] / [5.0, 5.0, 5.0] = [63.0 63.0 63.0]
>>> # input_x[1] = [63.0 63.0 63.0] / [7.0, 7.0, 7.0] = [9.0, 9.0, 9.0]
>>> indices = Tensor(np.array([[1, 0], [1, 1]]), mindspore.int32)
>>> updates = Tensor(np.array([[[1.0, 1.0, 1.0], [3.0, 3.0, 3.0]],
... [[5.0, 5.0, 5.0], [7.0, 7.0, 7.0]]]), mindspore.float32)
>>> output = ops.scatter_div(input_x, indices, updates)
>>> print(output)
[[35. 35. 35.]
[ 9. 9. 9.]]
"""
return scatter_div_(input_x, indices, updates)
[docs]def scatter_update(input_x, indices, updates):
r"""
Updates tensor values by using input indices and value.
Using given values to update tensor value, along with the input indices.
for each `i, ..., j` in `indices.shape`:
.. math::
\text{input_x}[\text{indices}[i, ..., j], :] = \text{updates}[i, ..., j, :]
Inputs of `input_x` and `updates` comply with the implicit type conversion rules to make the data types consistent.
If they have different data types, the lower priority data type will be converted to
the relatively highest priority data type.
Args:
input_x (Parameter): The target tensor, with data type of Parameter.
indices (Tensor): The index of input tensor. With int32 or int64 data type.
If there are duplicates in indices, the order for updating is undefined.
updates (Tensor): The tensor to update the input tensor, has the same type as input,
and updates.shape = indices.shape + input_x.shape[1:].
Returns:
Tensor, has the same shape and type as `input_x`.
Raises:
TypeError: If `indices` is not an int32 or an int64.
ValueError: If the shape of `updates` is not equal to `indices.shape + input_x.shape[1:]`.
RuntimeError: If the data type of `input_x` and `updates` conversion of Parameter
is required when data type conversion of Parameter is not supported.
Supported Platforms:
``Ascend`` ``GPU`` ``CPU``
Examples:
>>> import mindspore
>>> import numpy as np
>>> from mindspore import Tensor, ops
>>> np_x = np.array([[-0.1, 0.3, 3.6], [0.4, 0.5, -3.2]])
>>> input_x = mindspore.Parameter(Tensor(np_x, mindspore.float32), name="x")
>>> indices = Tensor(np.array([0, 1]), mindspore.int32)
>>> np_updates = np.array([[2.0, 1.2, 1.0], [3.0, 1.2, 1.0]])
>>> updates = Tensor(np_updates, mindspore.float32)
>>> output = ops.scatter_update(input_x, indices, updates)
>>> print(output)
[[2. 1.2 1.]
[3. 1.2 1.]]
"""
return scatter_update_(input_x, indices, updates)
[docs]def scatter_nd_add(input_x, indices, updates, use_locking=False):
r"""
Applies sparse addition to individual values or slices in a tensor.
Using given values to update tensor value through the add operation, along with the input indices.
This operation outputs the `input_x` after the update is done, which makes it convenient to use the updated value.
`input_x` has rank P and `indices` has rank Q where `Q >= 2`.
`indices` has shape :math:`(i_0, i_1, ..., i_{Q-2}, N)` where `N <= P`.
The last dimension of `indices` (with length `N` ) indicates slices along the `N` th dimension of `input_x`.
`updates` is a tensor of rank `Q-1+P-N`. Its shape is:
:math:`(i_0, i_1, ..., i_{Q-2}, x\_shape_N, ..., x\_shape_{P-1})`.
Args:
input_x (Parameter): The target tensor, with data type of Parameter.
indices (Tensor): The index to do min operation whose data type must be mindspore.int32 or mindspore.int64.
The rank of indices must be at least 2 and `indices.shape[-1] <= len(shape)`.
updates (Tensor): The tensor doing the addition operation with `input_x`,
the data type is same as `input_x`, the shape is `indices.shape[:-1] + x.shape[indices.shape[-1]:]`.
use_locking (bool): Whether to protect the assignment by a lock. Default: ``False`` .
Returns:
Tensor, the updated `input_x`, has the same shape and type as `input_x`.
Raises:
TypeError: If the dtype of `use_locking` is not bool.
TypeError: If the dtype of `indices` is not int32 or int64.
TypeError: If dtype of `input_x` and `updates` are not the same.
ValueError: If the shape of `updates` is not equal to `indices.shape[:-1] + x.shape[indices.shape[-1]:]`.
RuntimeError: If the data type of `input_x` and `updates` conversion of Parameter
is required when data type conversion of Parameter is not supported.
Supported Platforms:
``Ascend`` ``GPU`` ``CPU``
Examples:
>>> import mindspore
>>> import numpy as np
>>> from mindspore import Tensor, ops, Parameter
>>> input_x = Parameter(Tensor(np.array([1, 2, 3, 4, 5, 6, 7, 8]), mindspore.float32), name="x")
>>> indices = Tensor(np.array([[2], [4], [1], [7]]), mindspore.int32)
>>> updates = Tensor(np.array([6, 7, 8, 9]), mindspore.float32)
>>> output = ops.scatter_nd_add(input_x, indices, updates, False)
>>> print(output)
[ 1. 10. 9. 4. 12. 6. 7. 17.]
>>> input_x = Parameter(Tensor(np.zeros((4, 4, 4)), mindspore.int32))
>>> indices = Tensor(np.array([[0], [2]]), mindspore.int32)
>>> updates = Tensor(np.array([[[1, 1, 1, 1], [2, 2, 2, 2], [3, 3, 3, 3], [4, 4, 4, 4]],
... [[5, 5, 5, 5], [6, 6, 6, 6], [7, 7, 7, 7], [8, 8, 8, 8]]]), mindspore.int32)
>>> output = ops.scatter_nd_add(input_x, indices, updates, False)
>>> print(output)
[[[1 1 1 1]
[2 2 2 2]
[3 3 3 3]
[4 4 4 4]]
[[0 0 0 0]
[0 0 0 0]
[0 0 0 0]
[0 0 0 0]]
[[5 5 5 5]
[6 6 6 6]
[7 7 7 7]
[8 8 8 8]]
[[0 0 0 0]
[0 0 0 0]
[0 0 0 0]
[0 0 0 0]]]
"""
scatter_nd_add_inner = _get_cache_prim(P.ScatterNdAdd)(use_locking)
return scatter_nd_add_inner(input_x, indices, updates)
[docs]def scatter_nd_sub(input_x, indices, updates, use_locking=False):
r"""
Applies sparse subtraction to individual values or slices in a tensor.
Using given values to update tensor value through the subtraction operation, along with the input indices.
This operation outputs the `input_x` after the update is done, which makes it convenient to use the updated value.
`input_x` has rank P and `indices` has rank Q where `Q >= 2`.
`indices` has shape :math:`(i_0, i_1, ..., i_{Q-2}, N)` where `N <= P`.
The last dimension of `indices` (with length `N` ) indicates slices along the `N` th dimension of `input_x`.
`updates` is a tensor of rank `Q-1+P-N`. Its shape is:
:math:`(i_0, i_1, ..., i_{Q-2}, x\_shape_N, ..., x\_shape_{P-1})`.
Args:
input_x (Parameter): The target tensor, with data type of Parameter.
indices (Tensor): The index of input tensor, with int32 or int64 data type.
The rank of indices must be at least 2 and `indices.shape[-1] <= len(shape)`.
updates (Tensor): The tensor doing the subtraction operation with `input_x`, has the same type as input.
The shape is `indices.shape[:-1] + x.shape[indices.shape[-1]:]`.
use_locking (bool): Whether to protect the assignment by a lock. Default: ``False`` .
Returns:
Tensor, has the same shape and type as `input_x`.
Raises:
TypeError: If the dtype of `use_locking` is not bool.
TypeError: If the dtype of `indices` is not int32 or int64.
TypeError: If dtype of `input_x` and `updates` are not the same.
ValueError: If the shape of `updates` is not equal to `indices.shape[:-1] + x.shape[indices.shape[-1]:]`.
RuntimeError: If the data type of `input_x` and `updates` conversion of Parameter
is required when data type conversion of Parameter is not supported.
Supported Platforms:
``Ascend`` ``GPU`` ``CPU``
Examples:
>>> import mindspore
>>> import numpy as np
>>> from mindspore import Tensor, ops, Parameter
>>> input_x = Parameter(Tensor(np.array([1, 2, 3, 4, 5, 6, 7, 8]), mindspore.float32), name="x")
>>> indices = Tensor(np.array([[2], [4], [1], [7]]), mindspore.int32)
>>> updates = Tensor(np.array([6, 7, 8, 9]), mindspore.float32)
>>> output = ops.scatter_nd_sub(input_x, indices, updates, False)
>>> print(output)
[ 1. -6. -3. 4. -2. 6. 7. -1.]
>>> input_x = Parameter(Tensor(np.zeros((4, 4, 4)), mindspore.int32))
>>> indices = Tensor(np.array([[0], [2]]), mindspore.int32)
>>> updates = Tensor(np.array([[[1, 1, 1, 1], [2, 2, 2, 2], [3, 3, 3, 3], [4, 4, 4, 4]],
... [[5, 5, 5, 5], [6, 6, 6, 6], [7, 7, 7, 7], [8, 8, 8, 8]]]), mindspore.int32)
>>> output = ops.scatter_nd_sub(input_x, indices, updates, False)
>>> print(output)
[[[-1 -1 -1 -1]
[-2 -2 -2 -2]
[-3 -3 -3 -3]
[-4 -4 -4 -4]]
[[ 0 0 0 0]
[ 0 0 0 0]
[ 0 0 0 0]
[ 0 0 0 0]]
[[-5 -5 -5 -5]
[-6 -6 -6 -6]
[-7 -7 -7 -7]
[-8 -8 -8 -8]]
[[ 0 0 0 0]
[ 0 0 0 0]
[ 0 0 0 0]
[ 0 0 0 0]]]
"""
scatter_nd_sub_inner = _get_cache_prim(P.ScatterNdSub)(use_locking)
return scatter_nd_sub_inner(input_x, indices, updates)
[docs]def scatter_nd_mul(input_x, indices, updates, use_locking=False):
r"""
Applies sparse multiplication to individual values or slices in a tensor.
Using given values to update parameter value through the multiplication operation, along with the input indices.
This operation outputs the `input_x` after the update is done, which makes it convenient to use the updated value.
`input_x` has rank P and `indices` has rank Q, where `Q >= 2`.
`indices` has shape :math:`(i_0, i_1, ..., i_{Q-2}, N)` where `N <= P`.
The last dimension of `indices` (with length `N` ) indicates slices along the `N` th dimension of `input_x`.
`updates` is a tensor of rank `Q-1+P-N`. Its shape is:
:math:`(i_0, i_1, ..., i_{Q-2}, x\_shape_N, ..., x\_shape_{P-1})`.
Args:
input_x (Parameter): Input parameter.
indices (Tensor): The index to do multiplication operation whose data type must be mindspore.int32 or
mindspore.int64. The rank of indices must be at least 2 and `indices.shape[-1] <= len(shape)`.
updates (Tensor): The tensor to do the multiplication operation with `input_x`.
The data type is same as `input_x`, and the shape is `indices.shape[:-1] + x.shape[indices.shape[-1]:]`.
use_locking (bool): Whether to protect the assignment by a lock. Default: ``False`` .
Returns:
Tensor, the updated `input_x`, has the same shape and type as `input_x`.
Raises:
TypeError: If the dtype of `use_locking` is not bool.
TypeError: If the dtype of `indices` is not int32 or int64.
TypeError: If dtype of `input_x` and `updates` are not the same.
ValueError: If the shape of `updates` is not equal to `indices.shape[:-1] + x.shape[indices.shape[-1]:]`.
RuntimeError: If the data type of `input_x` and `updates` conversion of Parameter
is required when data type conversion of Parameter is not supported.
Supported Platforms:
``GPU`` ``CPU``
Examples:
>>> import mindspore
>>> import numpy as np
>>> from mindspore import Tensor, ops, Parameter
>>> input_x = Parameter(Tensor(np.array([1, 2, 3, 4, 5, 6, 7, 8]), mindspore.float32), name="x")
>>> indices = Tensor(np.array([[2], [4], [1], [7]]), mindspore.int32)
>>> updates = Tensor(np.array([6, 7, 8, 9]), mindspore.float32)
>>> output = ops.scatter_nd_mul(input_x, indices, updates)
>>> print(output)
[ 1. 16. 18. 4. 35. 6. 7. 72.]
>>> input_x = Parameter(Tensor(np.ones((4, 4, 4)), mindspore.int32))
>>> indices = Tensor(np.array([[0], [2]]), mindspore.int32)
>>> updates = Tensor(np.array([[[1, 1, 1, 1], [2, 2, 2, 2], [3, 3, 3, 3], [4, 4, 4, 4]],
... [[5, 5, 5, 5], [6, 6, 6, 6], [7, 7, 7, 7], [8, 8, 8, 8]]]), mindspore.int32)
>>> output = ops.scatter_nd_mul(input_x, indices, updates)
>>> print(output)
[[[1 1 1 1]
[2 2 2 2]
[3 3 3 3]
[4 4 4 4]]
[[1 1 1 1]
[1 1 1 1]
[1 1 1 1]
[1 1 1 1]]
[[5 5 5 5]
[6 6 6 6]
[7 7 7 7]
[8 8 8 8]]
[[1 1 1 1]
[1 1 1 1]
[1 1 1 1]
[1 1 1 1]]]
"""
scatter_nd_mul_inner = _get_cache_prim(ScatterNdMul)(use_locking)
return scatter_nd_mul_inner(input_x, indices, updates)
[docs]def scatter_nd_div(input_x, indices, updates, use_locking=False):
r"""
Applying sparse division to individual values or slices in a tensor.
Using given values to update tensor value through the div operation, along with the input indices.
This operation outputs the `input_x` after the update is done, which makes it convenient to use the updated value.
`input_x` has rank P and `indices` has rank Q, where `Q >= 2`.
`indices` has shape :math:`(i_0, i_1, ..., i_{Q-2}, N)` where `N <= P`.
The last dimension of `indices` (with length `N` ) indicates slices along the `N` th dimension of `input_x`.
`updates` is a tensor of rank `Q-1+P-N`. Its shape is:
:math:`(i_0, i_1, ..., i_{Q-2}, x\_shape_N, ..., x\_shape_{P-1})`.
Args:
input_x (Parameter): The target tensor, with data type of Parameter.
indices (Tensor): The index to do div operation whose data type must be mindspore.int32 or mindspore.int64.
The rank of indices must be at least 2 and `indices.shape[-1] <= len(shape)`.
updates (Tensor): The tensor to do the div operation with `input_x`.
The data type is same as `input_x`, and the shape is `indices.shape[:-1] + x.shape[indices.shape[-1]:]`.
use_locking (bool): Whether to protect the assignment by a lock. Default: ``False`` .
Returns:
Tensor, the updated `input_x`, has the same shape and type as `input_x`.
Raises:
TypeError: If the dtype of `use_locking` is not bool.
TypeError: If the dtype of `indices` is not int32 or int64.
TypeError: If dtype of `input_x` and `updates` are not the same.
ValueError: If the shape of `updates` is not equal to `indices.shape[:-1] + x.shape[indices.shape[-1]:]`.
RuntimeError: If the data type of `input_x` and `updates` conversion of Parameter
is required when data type conversion of Parameter is not supported.
Supported Platforms:
``GPU`` ``CPU``
Examples:
>>> import mindspore
>>> import numpy as np
>>> from mindspore import Tensor, ops, Parameter
>>> input_x = Parameter(Tensor(np.array([1, 2, 3, 4, 5, 6, 7, 8]), mindspore.float32), name="x")
>>> indices = Tensor(np.array([[2], [4], [1], [7]]), mindspore.int32)
>>> updates = Tensor(np.array([6, 7, 8, 9]), mindspore.float32)
>>> output = ops.scatter_nd_div(input_x, indices, updates, False)
>>> print(output)
[1. 0.25 0.5 4. 0.71428573 6.
7. 0.8888889 ]
>>> input_x = Parameter(Tensor(np.ones((4, 4, 4)), mindspore.float32))
>>> indices = Tensor(np.array([[0], [2]]), mindspore.int32)
>>> updates = Tensor(np.array([[[1, 1, 1, 1], [2, 2, 2, 2], [3, 3, 3, 3], [4, 4, 4, 4]],
... [[5, 5, 5, 5], [6, 6, 6, 6], [7, 7, 7, 7], [8, 8, 8, 8]]]), mindspore.float32)
>>> output = ops.scatter_nd_div(input_x, indices, updates, False)
>>> print(output)
[[[1. 1. 1. 1. ]
[0.5 0.5 0.5 0.5 ]
[0.33333334 0.33333334 0.33333334 0.33333334]
[0.25 0.25 0.25 0.25 ]]
[[1. 1. 1. 1. ]
[1. 1. 1. 1. ]
[1. 1. 1. 1. ]
[1. 1. 1. 1. ]]
[[0.2 0.2 0.2 0.2 ]
[0.16666667 0.16666667 0.16666667 0.16666667]
[0.14285715 0.14285715 0.14285715 0.14285715]
[0.125 0.125 0.125 0.125 ]]
[[1. 1. 1. 1. ]
[1. 1. 1. 1. ]
[1. 1. 1. 1. ]
[1. 1. 1. 1. ]]]
"""
scatter_nd_div_inner = _get_cache_prim(P.ScatterNdDiv)(use_locking)
return scatter_nd_div_inner(input_x, indices, updates)
[docs]def scatter_nd_max(input_x, indices, updates, use_locking=False):
r"""
Applying sparse maximum to individual values or slices in a tensor.
Using given values to update parameter value through the max operation, along with the input indices.
This operation outputs the `input_x` after the update is done, which makes it convenient to use the updated value.
`input_x` has rank P and `indices` has rank Q where `Q >= 2`.
`indices` has shape :math:`(i_0, i_1, ..., i_{Q-2}, N)` where `N <= P`.
The last dimension of `indices` (with length `N` ) indicates slices along the `N` th dimension of `input_x`.
`updates` is a tensor of rank `Q-1+P-N`. Its shape is:
:math:`(i_0, i_1, ..., i_{Q-2}, x\_shape_N, ..., x\_shape_{P-1})`.
Args:
input_x (Parameter): The target tensor, with data type of Parameter.
indices (Tensor): The index to do maximum operation whose data type must be mindspore.int32 or mindspore.int64.
The rank of indices must be at least 2 and `indices.shape[-1] <= len(shape)`.
updates (Tensor): The tensor to do the max operation with `input_x`.
The data type is same as `input_x`, and the shape is `indices.shape[:-1] + x.shape[indices.shape[-1]:]`.
use_locking (bool): Whether to protect the assignment by a lock. Default: ``False`` .
Returns:
Tensor, the updated `input_x`, has the same shape and type as `input_x`.
Raises:
TypeError: If the dtype of `use_locking` is not bool.
TypeError: If the dtype of `indices` is not int32 or int64.
TypeError: If dtype of `input_x` and `updates` are not the same.
ValueError: If the shape of `updates` is not equal to `indices.shape[:-1] + x.shape[indices.shape[-1]:]`.
RuntimeError: If the data type of `input_x` and `updates` conversion of Parameter
is required when data type conversion of Parameter is not supported.
Supported Platforms:
``Ascend`` ``GPU`` ``CPU``
Examples:
>>> import mindspore
>>> import numpy as np
>>> from mindspore import Tensor, ops, Parameter
>>> input_x = Parameter(Tensor(np.array([1, 2, 3, 4, 5, 6, 7, 8]), mindspore.float32), name="x")
>>> indices = Tensor(np.array([[2], [4], [1], [7]]), mindspore.int32)
>>> updates = Tensor(np.array([6, 7, 8, 9]), mindspore.float32)
>>> output = ops.scatter_nd_max(input_x, indices, updates, False)
>>> print(output)
[1. 8. 6. 4. 7. 6. 7. 9.]
>>> input_x = Parameter(Tensor(np.ones((4, 4, 4)), mindspore.int32))
>>> indices = Tensor(np.array([[0], [2]]), mindspore.int32)
>>> updates = Tensor(np.array([[[1, 1, 1, 1], [2, 2, 2, 2], [3, 3, 3, 3], [4, 4, 4, 4]],
... [[5, 5, 5, 5], [6, 6, 6, 6], [7, 7, 7, 7], [8, 8, 8, 8]]]), mindspore.int32)
>>> output = ops.scatter_nd_max(input_x, indices, updates, False)
>>> print(output)
[[[1 1 1 1]
[2 2 2 2]
[3 3 3 3]
[4 4 4 4]]
[[1 1 1 1]
[1 1 1 1]
[1 1 1 1]
[1 1 1 1]]
[[5 5 5 5]
[6 6 6 6]
[7 7 7 7]
[8 8 8 8]]
[[1 1 1 1]
[1 1 1 1]
[1 1 1 1]
[1 1 1 1]]]
"""
scatter_nd_max_inner = _get_cache_prim(ScatterNdMax)(use_locking)
return scatter_nd_max_inner(input_x, indices, updates)
[docs]def scatter_nd_min(input_x, indices, updates, use_locking=False):
r"""
Applying sparse minimum to individual values or slices in a tensor.
Using given values to update tensor value through the min operation, along with the input indices.
This operation outputs the `input_x` after the update is done, which makes it convenient to use the updated value.
`input_x` has rank P and `indices` has rank Q where `Q >= 2`.
`indices` has shape :math:`(i_0, i_1, ..., i_{Q-2}, N)` where `N <= P`.
The last dimension of `indices` (with length `N` ) indicates slices along the `N` th dimension of `input_x`.
`updates` is a tensor of rank `Q-1+P-N`. Its shape is:
:math:`(i_0, i_1, ..., i_{Q-2}, x\_shape_N, ..., x\_shape_{P-1})`.
Args:
input_x (Parameter): The target tensor, with data type of Parameter.
indices (Tensor): The index to do min operation whose data type must be mindspore.int32 or mindspore.int64.
The rank of indices must be at least 2 and `indices.shape[-1] <= len(shape)`.
updates (Tensor): The tensor to do the min operation with `input_x`.
The data type is same as `input_x`, and the shape is `indices.shape[:-1] + x.shape[indices.shape[-1]:]`.
use_locking (bool): Whether to protect the assignment by a lock. Default: ``False`` .
Returns:
Tensor, the updated `input_x`, has the same shape and type as `input_x`.
Raises:
TypeError: If the dtype of `use_locking` is not bool.
TypeError: If the dtype of `indices` is not int32 or int64.
TypeError: If dtype of `input_x` and `updates` are not the same.
ValueError: If the shape of `updates` is not equal to `indices.shape[:-1] + x.shape[indices.shape[-1]:]`.
RuntimeError: If the data type of `input_x` and `updates` conversion of Parameter
is required when data type conversion of Parameter is not supported.
Supported Platforms:
``Ascend`` ``GPU`` ``CPU``
Examples:
>>> import mindspore
>>> import numpy as np
>>> from mindspore import Tensor, ops, Parameter
>>> input_x = Parameter(Tensor(np.ones(8) * 10, mindspore.float32), name="x")
>>> indices = Tensor(np.array([[2], [4], [1], [7]]), mindspore.int32)
>>> updates = Tensor(np.array([6, 7, 8, 9]), mindspore.float32)
>>> output = ops.scatter_nd_min(input_x, indices, updates, False)
>>> print(output)
[10. 8. 6. 10. 7. 10. 10. 9.]
>>> input_x = Parameter(Tensor(np.ones((4, 4, 4)) * 10, mindspore.int32))
>>> indices = Tensor(np.array([[0], [2]]), mindspore.int32)
>>> updates = Tensor(np.array([[[1, 1, 1, 1], [2, 2, 2, 2], [3, 3, 3, 3], [4, 4, 4, 4]],
... [[5, 5, 5, 5], [6, 6, 6, 6], [7, 7, 7, 7], [8, 8, 8, 8]]]), mindspore.int32)
>>> output = ops.scatter_nd_min(input_x, indices, updates, False)
>>> print(output)
[[[ 1 1 1 1]
[ 2 2 2 2]
[ 3 3 3 3]
[ 4 4 4 4]]
[[10 10 10 10]
[10 10 10 10]
[10 10 10 10]
[10 10 10 10]]
[[ 5 5 5 5]
[ 6 6 6 6]
[ 7 7 7 7]
[ 8 8 8 8]]
[[10 10 10 10]
[10 10 10 10]
[10 10 10 10]
[10 10 10 10]]]
"""
scatter_nd_min_inner = _get_cache_prim(P.ScatterNdMin)(use_locking)
return scatter_nd_min_inner(input_x, indices, updates)
[docs]def sort(input_x, axis=-1, descending=False):
r"""
Sorts the elements of the input tensor along the given dimension in the specified order.
Args:
input_x(Tensor): The input tensor to sort.
The shape is :math:`(N,*)` where :math:`*` means, any number of additional dimensions.
axis (int, optional): The dimension to sort along. Default: ``-1``, means the last dimension.
The Ascend backend only supports sorting the last dimension.
descending (bool, optional): Controls the sort order. If `descending` is True, the elements
are sorted in descending order, or else sorted in ascending order. Default: ``False`` .
.. warning::
Currently, the data types of float16, uint8, int8, int16, int32, int64 are well supported.
If use float32, it may cause loss of accuracy.
Returns:
- y1, a tensor whose values are the sorted values, with the same shape and data type as input.
- y2, a tensor that consists of the indices of the elements in the original input tensor.
Data type is int32.
Raises:
TypeError: If `axis` is not an int.
TypeError: If `descending` is not a bool.
TypeError: If dtype of `input_x` is neither float16, float32, uint8, int8, int16, int32, int64.
ValueError: If `axis` is not in range of [-len(input_x.shape), len(input_x.shape)).
Supported Platforms:
``Ascend`` ``GPU`` ``CPU``
Examples:
>>> import mindspore
>>> import numpy as np
>>> from mindspore import Tensor, ops
>>> x = Tensor(np.array([[8, 2, 1], [5, 9, 3], [4, 6, 7]]), mindspore.float16)
>>> output = ops.sort(x)
>>> # The output below is based on the Ascend platform.
>>> print(output)
(Tensor(shape=[3, 3], dtype=Float16, value=
[[ 1.0000e+00, 2.0000e+00, 8.0000e+00],
[ 3.0000e+00, 5.0000e+00, 9.0000e+00],
[ 4.0000e+00, 6.0000e+00, 7.0000e+00]]), Tensor(shape=[3, 3], dtype=Int32, value=
[[2, 1, 0],
[2, 0, 1],
[0, 1, 2]]))
"""
_sort = _get_cache_prim(P.Sort)(axis, descending)
return _sort(input_x)
[docs]def sort_ext(input, *, dim=-1, descending=False, stable=False):
r"""
Sorts the elements of the input tensor along the given dimension in the specified order.
.. warning::
Currently, the data types of float16, uint8, int8, int16, int32, int64 are well supported.
If use float32, it may cause loss of accuracy.
Args:
input(Tensor): The input tensor to sort.
The shape is :math:`(N,*)` where :math:`*` means, any number of additional dimensions.
Keyword Args:
dim (int, optional): The dimension to sort along. Default: ``-1``, means the last dimension.
descending (bool, optional): Controls the sort order. If `descending` is True, the elements
are sorted in descending order, or else sorted in ascending order. Default: ``False`` .
stable (bool, optional): Controls the sort order. If stable is True then the sorting routine
becomes stable, preserving the order of equivalent elements. Default: ``False`` .
Returns:
- y1, a tensor whose values are the sorted values, with the same shape and data type as input.
- y2, a tensor that consists of the indices of the elements in the original input tensor.
Data type is int64.
Raises:
TypeError: If `dim` is not an int.
TypeError: If `descending` is not a bool.
TypeError: If `input` not in float16, float32, uint8, int8, int16, int32, int64, bfloat16
TypeError: If `stable` is not a bool.
ValueError: If `dim` is not in range of [-len(input_x.shape), len(input_x.shape)).
Supported Platforms:
``Ascend``
Examples:
>>> import mindspore
>>> import numpy as np
>>> from mindspore import Tensor, ops
>>> x = Tensor(np.array([[8, 2, 1], [5, 9, 3], [4, 6, 7]]), mindspore.float16)
>>> output = ops.function.array_func.sort_ext(x)
>>> # The output below is based on the Ascend platform.
>>> print(output)
(Tensor(shape=[3, 3], dtype=Float16, value=
[[ 1.0000e+00, 2.0000e+00, 8.0000e+00],
[ 3.0000e+00, 5.0000e+00, 9.0000e+00],
[ 4.0000e+00, 6.0000e+00, 7.0000e+00]]), Tensor(shape=[3, 3], dtype=Int64, value=
[[2, 1, 0],
[2, 0, 1],
[0, 1, 2]]))
"""
return sort_ext_(input, dim, descending, stable)
[docs]def argsort(input, axis=-1, descending=False):
r"""
Sorts the input tensor along the given dimension in specified order and return the sorted indices.
Args:
input(Tensor): The input tensor to sort.
axis (int): The axis to sort along. Default: ``-1`` , means the last dimension.
The Ascend backend only supports sorting the last dimension.
descending (bool): The sort order. If `descending` is True then the elements
are sorted in descending order by value. Otherwise sort in ascending order. Default: ``False`` .
Returns:
Tensor, the indices of sorted input tensor. Data type is int32.
Supported Platforms:
``Ascend`` ``GPU`` ``CPU``
Examples:
>>> import mindspore
>>> import numpy as np
>>> from mindspore import Tensor, ops
>>> x = Tensor(np.array([[8, 2, 1], [5, 9, 3], [4, 6, 7]]), mindspore.float16)
>>> sort = ops.argsort(x)
>>> print(sort)
[[2 1 0]
[2 0 1]
[0 1 2]]
"""
_sort = _get_cache_prim(P.Sort)(axis, descending)
_, arg_sort = _sort(input)
return arg_sort
[docs]def gather_elements(input, dim, index):
"""
Gathers elements along an axis specified by dim.
For a 3-D tensor, the output is:
.. code-block::
output[i][j][k] = x[index[i][j][k]][j][k] # if dim == 0
output[i][j][k] = x[i][index[i][j][k]][k] # if dim == 1
output[i][j][k] = x[i][j][index[i][j][k]] # if dim == 2
`input` and `index` have the same length of dimensions, and `index.shape[axis] <= input.shape[axis]`
where axis goes through all dimensions of `input` except `dim`.
.. warning::
On Ascend, the behavior is unpredictable in the following cases:
- the value of `index` is not in the range `[-input.shape[dim], input.shape[dim])` in forward;
- the value of `index` is not in the range `[0, input.shape[dim])` in backward.
Args:
input (Tensor): The input tensor.
dim (int): The axis along which to index. It must be int32 or int64. The value range is `[-input.ndim,
input.ndim)`.
index (Tensor): The indices of elements to gather. It can be one of the following data types:
int32, int64. The value range of each index element is `[-input.shape(dim), input.shape(dim))`.
Returns:
Tensor, has the same shape as `index` and has the same data type with `input`.
Raises:
TypeError: If dtype of `dim` or `index` is neither int32 nor int64.
ValueError: If length of shape of `input` is not equal to length of shape of `index`.
ValueError: If the size of the dimension except `dim` in `input` is less than size in `index`.
ValueError: If the value of `dim` is not in the expected range.
Supported Platforms:
``Ascend`` ``GPU`` ``CPU``
Examples:
>>> import numpy as np
>>> import mindspore
>>> from mindspore import Tensor
>>> x = Tensor(np.array([[1, 2], [3, 4]]), mindspore.int32)
>>> index = Tensor(np.array([[0, 0], [1, 0]]), mindspore.int32)
>>> dim = 1
>>> output = mindspore.ops.gather_elements(x, dim, index)
>>> print(output)
[[1 1]
[4 3]]
"""
return gather_d_(input, dim, index)
[docs]def tensor_scatter_add(input_x, indices, updates):
r"""
Creates a new tensor by adding the values from the positions in `input_x` indicated by
`indices`, with values from `updates`. When multiple values are given for the same
index, the updated result will be the sum of all values. This operation is almost
equivalent to using ScatterNdAdd, except that the updates are applied on output `Tensor`
instead of input `Parameter`.
The last axis of `indices` is the depth of each index vectors. For each index vector,
there must be a corresponding value in `updates`. The shape of `updates` should be
equal to the shape of `input_x[indices]`. For more details, see Examples.
.. math::
output\left [indices \right ] = input\_x + update
Note:
- On GPU, if some values of the `indices` are out of bound, instead of raising an index error,
the corresponding `updates` will not be updated to self tensor.
- On CPU, if some values of the `indices` are out of bound, raising an index error.
- On Ascend, out of bound checking is not supported, if some values of the `indices` are out of bound,
unknown errors may be caused.
Args:
input_x (Tensor): The input tensor. The dimension of input_x must be no less than indices.shape[-1].
indices (Tensor): The index of input tensor whose data type is int32 or int64.
The rank must be at least 2.
updates (Tensor): The tensor to update the input tensor, has the same type as input,
and updates. And the shape should be
equal to :math:`indices.shape[:-1] + input\_x.shape[indices.shape[-1]:]`.
Returns:
Tensor, has the same shape and type as `input_x`.
Raises:
TypeError: If dtype of `indices` is neither int32 nor int64.
ValueError: If length of shape of `input_x` is less than the last dimension of shape of `indices`.
RuntimeError: If a value of `indices` is not in `input_x` on CPU backend.
Supported Platforms:
``Ascend`` ``GPU`` ``CPU``
Examples:
>>> import mindspore
>>> import numpy as np
>>> from mindspore import Tensor, nn
>>> from mindspore import ops
>>> input_x = Tensor(np.array([[-0.1, 0.3, 3.6], [0.4, 0.5, -3.2]]), mindspore.float32)
>>> indices = Tensor(np.array([[0, 0], [0, 0]]), mindspore.int32)
>>> updates = Tensor(np.array([1.0, 2.2]), mindspore.float32)
>>> output = ops.tensor_scatter_add(input_x, indices, updates)
>>> print(output)
[[ 3.1 0.3 3.6]
[ 0.4 0.5 -3.2]]
"""
return tensor_scatter_add_(input_x, indices, updates)
[docs]def tensor_scatter_sub(input_x, indices, updates):
r"""
Creates a new tensor by subtracting the values from the positions in `input_x` indicated by
`indices`, with values from `updates`. When multiple values are provided for the same
index, the result of the update will be to subtract these values respectively. This operation is almost
equivalent to using :class:`mindspore.ops.ScatterNdSub` , except that the updates are applied on output `Tensor`
instead of input `Parameter`.
The last axis of `indices` is the depth of each index vectors. For each index vector,
there must be a corresponding value in `updates`. The shape of `updates` should be
equal to the shape of `input_x[indices]`. For more details, see Examples.
.. math::
output[indices] = input\_x - update
Note:
On GPU, if some values of the `indices` are out of bound, instead of raising an index error,
the corresponding `updates` will not be updated to self tensor. On CPU, if some values of
the `indices` are out of bound, raising an index error. On Ascend, out of bound checking is
not supported, if some values of the `indices` are out of bound, unknown errors may be caused.
Args:
input_x (Tensor): The input tensor. The dimension of input_x must be no less than indices.shape[-1].
indices (Tensor): The index of input tensor whose data type is int32 or int64.
The rank must be at least 2.
updates (Tensor): The tensor to update the input tensor, has the same type as `input_x`,
and the shape of `updates` should be equal to indices.shape[:-1] + input_x.shape[indices.shape[-1]:].
Returns:
Tensor, has the same shape and type as `input_x`.
Raises:
TypeError: If dtype of `indices` is neither int32 nor int64.
ValueError: If length of shape of `input_x` is less than the last dimension of shape of `indices`.
RuntimeError: If a value of `indices` is not in `input_x`.
Supported Platforms:
``Ascend`` ``GPU`` ``CPU``
Examples:
>>> import mindspore
>>> import numpy as np
>>> from mindspore import Tensor
>>> from mindspore import ops
>>> input_x = Tensor(np.array([[-0.1, 0.3, 3.6], [0.4, 0.5, -3.2]]), mindspore.float32)
>>> indices = Tensor(np.array([[0, 0], [0, 0]]), mindspore.int32)
>>> updates = Tensor(np.array([1.0, 2.2]), mindspore.float32)
>>> output = ops.tensor_scatter_sub(input_x, indices, updates)
>>> print(output)
[[-3.3000002 0.3 3.6 ]
[ 0.4 0.5 -3.2 ]]
"""
return tensor_scatter_sub_(input_x, indices, updates)
[docs]def tensor_scatter_max(input_x, indices, updates):
r"""
By comparing the value at the position indicated by `indices` in `input_x` with the value in the `updates`,
the value at the index will eventually be equal to the largest one to create a new tensor.
The last axis of the index is the depth of each index vector. For each index vector,
there must be a corresponding value in `updates`. The shape of `updates` should be
equal to the shape of input_x[indices].
.. math::
output\left [indices \right ] = \max(input\_x, update)
Note:
- On GPU, if some values of the `indices` are out of bound, instead of raising an index error,
the corresponding `updates` will not be updated to self tensor.
- On CPU, if some values of the `indices` are out of bound, raising an index error.
- On Ascend, out of bound checking is not supported, if some values of the `indices` are out of bound,
unknown errors may be caused.
Args:
input_x (Tensor): The input tensor. The dimension of `input_x` must be no less than indices.shape[-1].
indices (Tensor): The index of input tensor whose data type must be int32 or int64.
The rank must be at least 2.
updates (Tensor): The tensor to update the `input_x` tensor, has the same type as input,
and updates.shape should be equal to :math:`indices.shape[:-1] + input\_x.shape[indices.shape[-1]:]`.
Returns:
Tensor, has the same shape and type as `input_x`.
Raises:
TypeError: If dtype of `indices` is neither int32 nor int64.
ValueError: If length of shape of `input_x` is less than the last dimension of shape of `indices`.
RuntimeError: If a value of `indices` is not in `input_x` on CPU backend.
Supported Platforms:
``GPU`` ``CPU``
Examples:
>>> import mindspore
>>> import numpy as np
>>> from mindspore import Tensor, ops
>>> input_x = Tensor(np.array([[-0.1, 0.3, 3.6], [0.4, 0.5, -3.2]]), mindspore.float32)
>>> indices = Tensor(np.array([[0, 0], [0, 0]]), mindspore.int32)
>>> updates = Tensor(np.array([1.0, 2.2]), mindspore.float32)
>>> # Next, demonstrate the approximate operation process of this operator:
>>> # 1, indices[0] = [0, 0], indices[1] = [0, 0]
>>> # 2, And input_x[0, 0] = -0.1
>>> # 3, So input_x[indices] = [-0.1, -0.1]
>>> # 4, Satisfy the above formula: input_x[indices].shape=(2) == updates.shape=(2)
>>> output = ops.tensor_scatter_max(input_x, indices, updates)
>>> # 5, Perform the max operation for the first time:
>>> # first_input_x = Max(input_x[0][0], updates[0]) = [[1.0, 0.3, 3.6], [0.4, 0.5, -3.2]]
>>> # 6, Perform the max operation for the second time:
>>> # second_input_x = Max(input_x[0][0], updates[1]) = [[2.2, 0.3, 3.6], [0.4, 0.5, -3.2]]
>>> print(output)
[[ 2.2 0.3 3.6]
[ 0.4 0.5 -3.2]]
"""
return tensor_scatter_max_(input_x, indices, updates)
[docs]def tensor_scatter_min(input_x, indices, updates):
r"""
By comparing the value at the position indicated by `indices` in `input_x` with the value in the `updates`,
the value at the index will eventually be equal to the smallest one to create a new tensor.
The last axis of the index is the depth of each index vector. For each index vector,
there must be a corresponding value in `updates`. The shape of `updates` should be
equal to the shape of `input_x[indices]`. For more details, see case below.
.. math::
output\left [indices \right ] = \min(input\_x, update)
Note:
- On GPU, if some values of the `indices` are out of bound, instead of raising an index error,
the corresponding `updates` will not be updated to self tensor.
- On CPU, if some values of the `indices` are out of bound, raising an index error.
- On Ascend, out of bound checking is not supported, if some values of the `indices` are out of bound,
unknown errors may be caused.
Args:
input_x (Tensor): The input tensor. The dimension of `input_x` must be no less than indices.shape[-1].
indices (Tensor): The index of input tensor whose data type is int32 or int64.
The rank must be at least 2.
updates (Tensor): The tensor to update the input tensor, has the same type as `input_x`
And the shape of `updates` should be
equal to :math:`indices.shape[:-1] + input\_x.shape[indices.shape[-1]:]`.
Returns:
Tensor, has the same shape and type as `input_x`.
Raises:
TypeError: If dtype of `indices` is neither int32 nor int64.
ValueError: If length of shape of `input_x` is less than the last dimension of shape of `indices`.
RuntimeError: If a value of `indices` is not in `input_x` on CPU backend.
Supported Platforms:
``Ascend`` ``GPU`` ``CPU``
Examples:
>>> import mindspore
>>> import numpy as np
>>> from mindspore import Tensor
>>> from mindspore import ops
>>> input_x = Tensor(np.array([[-0.1, 0.3, 3.6], [0.4, 0.5, -3.2]]), mindspore.float32)
>>> indices = Tensor(np.array([[0, 0], [0, 0]]), mindspore.int32)
>>> updates = Tensor(np.array([1.0, 2.2]), mindspore.float32)
>>> output = ops.tensor_scatter_min(input_x, indices, updates)
>>> print(output)
[[ -0.1 0.3 3.6]
[ 0.4 0.5 -3.2]]
"""
return tensor_scatter_min_(input_x, indices, updates)
[docs]def tensor_scatter_elements(input_x, indices, updates, axis=0, reduction="none"):
"""
Write all elements in `updates` to the index specified by `indices` in `input_x` according to the reduction
operation specified by `reduction`.
`axis` controls the direction of the scatter operation.
`tensor_scatter_elements` takes three inputs `input_x`, `updates` and `indices` of the same rank r >= 1.
For a 3-D tensor, the output is:
.. code-block::
output[indices[i][j][k]][j][k] = updates[i][j][k] # if axis == 0, reduction == "none"
output[i][indices[i][j][k]][k] += updates[i][j][k] # if axis == 1, reduction == "add"
output[i][j][indices[i][j][k]] = updates[i][j][k] # if axis == 2, reduction == "none"
.. warning::
- The order in which updates are applied is nondeterministic, meaning that if there are multiple index vectors
in `indices` that correspond to the same position, the value of that position in the output will be
nondeterministic.
- On Ascend, the reduction only support set to "none" for now.
- On Ascend, the data type of `input_x` must be float16 or float32.
- This is an experimental API that is subject to change or deletion.
Note:
If some values of the `indices` exceed the upper or lower bounds of the index of `input_x`, instead of raising
an index error, the corresponding `updates` will not be updated to `input_x`.
The backward is supported only for the case `updates.shape == indices.shape`.
Args:
input_x (Tensor): The target tensor. The rank must be at least 1.
indices (Tensor): The index of `input_x` to do scatter operation whose data type must be mindspore.int32 or
mindspore.int64. Same rank as `input_x`. And accepted range is [-s, s) where s is the size along axis.
updates (Tensor): The tensor doing the scatter operation with `input_x`.
axis (int): Which axis to scatter. Accepted range is [-r, r) where r = rank(input_x). Default: ``0``.
reduction (str): Which reduction operation to scatter, supports ``"none"`` , ``"add"`` . Default: ``"none"``.
When `reduction` is set to ``"none"``, `updates` will be assigned to `input_x` according to `indices`.
When `reduction` is set to ``"add"``, `updates` will be added to `input_x` according to `indices`.
Returns:
Tensor, has the same shape and type as `input_x`.
Raises:
TypeError: If `indices` is neither int32 nor int64.
ValueError: If anyone of the rank among `input_x`, `indices` and `updates` less than 1.
ValueError: If the rank of `updates` is not equal to the rank of `input_x`.
RuntimeError: If the data type of `input_x` and `updates` conversion of Parameter
is required when data type conversion of Parameter is not supported.
Supported Platforms:
``Ascend`` ``GPU`` ``CPU``
Examples:
>>> import mindspore
>>> from mindspore import Tensor, ops
>>> from mindspore import Parameter
>>> import numpy as np
>>> input_x = Parameter(Tensor(np.array([[1, 2, 3, 4, 5]]), mindspore.int32), name="x")
>>> indices = Tensor(np.array([[2, 4]]), mindspore.int32)
>>> updates = Tensor(np.array([[8, 8]]), mindspore.int32)
>>> axis = 1
>>> reduction = "none"
>>> output = ops.tensor_scatter_elements(input_x, indices, updates, axis, reduction)
>>> print(output)
[[1 2 8 4 8]]
>>> input_x = Parameter(Tensor(np.array([[1, 2, 3], [4, 5, 6], [7, 8, 9]]), mindspore.int32), name="x")
>>> indices = Tensor(np.array([[1, -1, 2], [0, 2, 1]]), mindspore.int32)
>>> updates = Tensor(np.array([[1, 2, 2], [4, 5, 8]]), mindspore.int32)
>>> axis = 0
>>> reduction = "add"
>>> output = ops.tensor_scatter_elements(input_x, indices, updates, axis, reduction)
>>> print(output)
[[ 5 2 3]
[ 5 5 14]
[ 7 15 11]]
"""
return tensor_scatter_elements_ext(input_x, indices, updates, axis, reduction)
[docs]def scatter(input, axis, index, src):
"""
Update the value in `src` to `input` according to the specified index.
Refer to :func:`mindspore.ops.tensor_scatter_elements` for more details.
.. note::
The backward is supported only for the case `src.shape == index.shape`.
Args:
input (Tensor): The target tensor. The rank of `input` must be at least 1.
axis (int): Which axis to scatter. Accepted range is [-r, r) where r = rank(input).
index (Tensor): The index to do update operation whose data must be positive number with type of mindspore.int32
or mindspore.int64. Same rank as `input` . And accepted range is [-s, s) where s is the size along axis.
src (Tensor, float): The data doing the update operation with `input`. Can be a tensor with the same data type
as `input` or a float number to scatter.
Returns:
The backward is supported only for the case `src.shape == index.shape` when `src` is a tensor.
Raises:
TypeError: If `index` is neither int32 nor int64.
ValueError: If rank of any of `input` , `index` and `src` less than 1.
ValueError: If the rank of `src` is not equal to the rank of `input` .
TypeError: If the data type of `input` and `src` have different dtypes.
RuntimeError: If `index` has negative elements.
Supported Platforms:
``Ascend`` ``GPU`` ``CPU``
Examples:
>>> import numpy as np
>>> import mindspore as ms
>>> from mindspore import Tensor, ops
>>> input = Tensor(np.array([[1, 2, 3, 4, 5]]), dtype=ms.float32)
>>> src = Tensor(np.array([[8, 8]]), dtype=ms.float32)
>>> index = Tensor(np.array([[2, 4]]), dtype=ms.int64)
>>> out = ops.scatter(input=input, axis=1, index=index, src=src)
>>> print(out)
[[1. 2. 8. 4. 8.]]
>>> input = Tensor(np.zeros((5, 5)), dtype=ms.float32)
>>> src = Tensor(np.array([[1, 2, 3], [4, 5, 6], [7, 8, 9]]), dtype=ms.float32)
>>> index = Tensor(np.array([[0, 0, 0], [2, 2, 2], [4, 4, 4]]), dtype=ms.int64)
>>> out = ops.scatter(input=input, axis=0, index=index, src=src)
>>> print(out)
[[1. 2. 3. 0. 0.]
[0. 0. 0. 0. 0.]
[4. 5. 6. 0. 0.]
[0. 0. 0. 0. 0.]
[7. 8. 9. 0. 0.]]
>>> input = Tensor(np.zeros((5, 5)), dtype=ms.float32)
>>> src = Tensor(np.array([[1, 2, 3], [4, 5, 6], [7, 8, 9]]), dtype=ms.float32)
>>> index = Tensor(np.array([[0, 2, 4], [0, 2, 4], [0, 2, 4]]), dtype=ms.int64)
>>> out = ops.scatter(input=input, axis=1, index=index, src=src)
>>> print(out)
[[1. 0. 2. 0. 3.]
[4. 0. 5. 0. 6.]
[7. 0. 8. 0. 9.]
[0. 0. 0. 0. 0.]
[0. 0. 0. 0. 0.]]
"""
if isinstance(src, Tensor):
return scatter_(input, axis, index, src)
return scatter_value_(input, axis, index, src)
[docs]def scatter_add_ext(input, dim, index, src):
"""
Add all elements in `src` to the index specified by `index` to `input` along dimension specified by `dim`.
It takes three inputs `input`, `src` and `index` of the same rank r >= 1.
For a 3-D tensor, the operation updates input as follows:
.. code-block::
input[index[i][j][k]][j][k] += src[i][j][k] # if dim == 0
input[i][index[i][j][k]][k] += src[i][j][k] # if dim == 1
input[i][j][index[i][j][k]] += src[i][j][k] # if dim == 2
Args:
input (Tensor): The target tensor. The rank must be at least 1.
dim (int): Which dim to scatter. Accepted range is [-r, r) where r = rank(`input`). Default: ``0``.
index (Tensor): The index of `input` to do scatter operation whose data type must be mindspore.int32 or
mindspore.int64. Same rank as `input`. Except for the dimension specified by `dim`,
the size of each dimension of `index` must be less than or equal to the size of
the corresponding dimension of `input`.
src (Tensor): The tensor doing the scatter operation with `input`, has the same type as `input` and
the size of each dimension must be greater than or equal to that of `index`.
Returns:
Tensor, has the same shape and type as `input`.
Raises:
TypeError: If `index` is neither int32 nor int64.
ValueError: If anyone of the rank among `input`, `index` and `src` less than 1.
ValueError: If the rank of `input`, `index` and `src` is not the same.
ValueError: If, outside dimension `dim`, the size of any dimension of `index` is greater than the size of
the corresponding dimension of `input` .
ValueError: If the size of any dimension of `src` is less than that of `index`.
Supported Platforms:
``Ascend``
Examples:
>>> import numpy as np
>>> import mindspore as ms
>>> from mindspore import Tensor, ops
>>> input = Tensor(np.array([[1, 2, 3, 4, 5]]), dtype=ms.float32)
>>> src = Tensor(np.array([[8, 8]]), dtype=ms.float32)
>>> index = Tensor(np.array([[2, 4]]), dtype=ms.int64)
>>> out = ops.function.array_func.scatter_add_ext(input=input, dim=1, index=index, src=src)
>>> print(out)
[[1. 2. 11. 4. 13.]]
>>> input = Tensor(np.zeros((5, 5)), dtype=ms.float32)
>>> src = Tensor(np.array([[1, 2, 3], [4, 5, 6], [7, 8, 9]]), dtype=ms.float32)
>>> index = Tensor(np.array([[0, 0, 0], [2, 2, 2], [4, 4, 4]]), dtype=ms.int64)
>>> out = ops.function.array_func.scatter_add_ext(input=input, dim=0, index=index, src=src)
>>> print(out)
[[1. 2. 3. 0. 0.]
[0. 0. 0. 0. 0.]
[4. 5. 6. 0. 0.]
[0. 0. 0. 0. 0.]
[7. 8. 9. 0. 0.]]
>>> input = Tensor(np.zeros((5, 5)), dtype=ms.float32)
>>> src = Tensor(np.array([[1, 2, 3], [4, 5, 6], [7, 8, 9]]), dtype=ms.float32)
>>> index = Tensor(np.array([[0, 2, 4], [0, 2, 4], [0, 2, 4]]), dtype=ms.int64)
>>> out = ops.function.array_func.scatter_add_ext(input=input, dim=1, index=index, src=src)
>>> print(out)
[[1. 0. 2. 0. 3.]
[4. 0. 5. 0. 6.]
[7. 0. 8. 0. 9.]
[0. 0. 0. 0. 0.]
[0. 0. 0. 0. 0.]]
"""
return scatter_add_ext_op(input, dim, index, src)
def _get_slice_scatter_const(x_shape, axis, start, end, step):
r"""
Calculate the rank of input, embedded dimensions and index.
"""
x_rank = len(x_shape)
axis = axis if axis >= 0 else axis + x_rank
start = start if start is not None else 0
start = start if start >= 0 else start + x_rank
end = end if end is not None else x_shape[axis]
end = end if end >= 0 else end + x_shape[axis]
end = end if end < x_shape[axis] else x_shape[axis]
index = list(builtins.range(start, end, step))
return x_rank, index, axis
[docs]def slice_scatter(input, src, axis=0, start=None, end=None, step=1):
r"""
Slice the input Tensor in the specified dimension and overlay the slice results with the source Tensor.
The `input` is sliced along the specified dimension. The start position of the slice is `start` ,
the end position is `end` , and the step size is `step` .
Then the slicing result is overwritten with `src` to get the output Tensor.
Args:
input (Tensor): The target Tensor.
src (Tensor): The source Tensor.
axis (int, optional): The dimension of `input` to be sliced. Default: ``0`` .
start (int, optional): The start index to slice in the specified dimension.
Default: ``None``, `start` is ``0`` .
end (int, optional): The end index to slice in the specified dimension.
Default: ``None``, `end` is the length of `input` in the specified dimension.
step (int, optional): Step size. Default: ``1``, the distance from the next slice element is ``1`` .
Returns:
Tensor after embedding, has the same shape and type as `input` .
Raises:
ValueError: The shape of `src` is not the same as the shape of `input` slice.
TypeError: If `input` is not a Tensor.
TypeError: If `src` is not a Tensor.
TypeError: If `axis` or `step` is not an integer.
TypeError: If `start` or `end` is not ``None`` or an integer.
Supported Platforms:
``Ascend`` ``GPU`` ``CPU``
Examples:
>>> import mindspore as ms
>>> a = ms.ops.zeros((4, 6))
>>> b = ms.ops.ones((4, 3))
>>> output = ms.ops.slice_scatter(a, b, axis=1, start=0, end=5, step=2)
>>> print(output)
[[1. 0. 1. 0. 1. 0.]
[1. 0. 1. 0. 1. 0.]
[1. 0. 1. 0. 1. 0.]
[1. 0. 1. 0. 1. 0.]]
"""
_check_is_tensor("input", input, "slice_scatter")
_check_is_tensor("src", src, "slice_scatter")
input_shape = input.shape
input_rank, index, axis = _get_slice_scatter_const(
input_shape, axis, start, end, step)
src_shape = src.shape
index_shape = input_shape[:axis] + (len(index),) + input_shape[axis + 1:]
index_tensor = ms.Tensor(index)
for _ in builtins.range(axis):
index_tensor = index_tensor.expand_dims(0)
if index_shape != src_shape:
raise ValueError(f"For slice_scatter, src shape should be equal to the slice size,"
f"but got src shape {src_shape} and slice shape {index_shape}")
for _ in builtins.range(input_rank - axis - 1):
index_tensor = index_tensor.expand_dims(-1)
index_tensor = index_tensor.broadcast_to(src.shape)
if index_tensor.dtype not in mstype.int_type:
index_tensor = index_tensor.astype(mstype.int64)
return tensor_scatter_elements(input, axis=axis, indices=index_tensor, updates=src)
[docs]def select_scatter(input, src, axis, index):
r"""
On the specified dimension `axis` of `input` , `src` is scattered into `input` on the specified `index` of `input` .
Args:
input (Tensor): The target Tensor.
src (Tensor): The source Tensor.
axis (int): The dimension of `input` to be embedded.
index (int): The location of scattering on the specified dimension.
Returns:
Tensor after embedding, has the same shape and type as `input` .
Raises:
ValueError: The shape of `src` is not the same as the shape scattered over `input` .
TypeError: If `input` is not a Tensor.
TypeError: If `src` is not a Tensor.
TypeError: If `axis` or `index` is not an integer.
Supported Platforms:
``Ascend`` ``GPU`` ``CPU``
Examples:
>>> import mindspore as ms
>>> a = ms.ops.zeros((2, 3, 3))
>>> b = ms.ops.ones((2, 3))
>>> output = ms.ops.select_scatter(a, b, axis=1, index=1)
>>> print(output)
[[[0. 0. 0.]
[1. 1. 1.]
[0. 0. 0.]]
[[0. 0. 0.]
[1. 1. 1.]
[0. 0. 0.]]]
"""
_check_is_tensor("input", input, "select_scatter")
_check_is_tensor("src", src, "select_scatter")
src = src.expand_dims(axis=axis)
x_rank = input.ndim
axis = axis if axis >= 0 else axis + x_rank
index = index if index >= 0 else index + input.shape[axis]
return slice_scatter(input, src, axis, start=index, end=index + 1)
[docs]def space_to_batch_nd(input_x, block_size, paddings):
r"""
Divides a tensor's spatial dimensions into blocks and combines the block sizes with the original batch.
This operation will divide spatial dimensions into blocks with `block_size`,
and after division, the output tensor's spatial dimension is the corresponding number of blocks.
The output tensor's batch dimension is the product of the original batch and the product of `block_size`.
Before division, the spatial dimensions of the input are zero padded according to paddings if necessary.
Assume input shape is :math:`(n, c_1, ... c_k, w_1, ..., w_M)`, then the shape of the output tensor will be
:math:`(n', c_1, ... c_k, w'_1, ..., w'_M)`, where
.. math::
\begin{array}{ll} \\
n' = n*(block\_size[0] * ... * block\_size[M]) \\
w'_i = (w_i + paddings[i][0] + paddings[i][1])//block\_size[i]
\end{array}
Args:
input_x (Tensor): The input tensor. It must be a 4-D tensor on Ascend.
block_size (Union[list(int), tuple(int), int]): The block size of dividing block with all value greater
than 1. If `block_size` is a tuple or list, the length of `block_size` is M corresponding to the
number of spatial dimensions. If `block_size` is an int, the block size of M dimensions are the same,
equal to `block_size`. M must be 2 on Ascend.
paddings (Union[tuple, list]): The padding values for spatial dimensions, containing M subtraction list.
Each contains 2 integer values. All values must be greater than 0.
`paddings[i]` specifies the paddings for the spatial dimension i,
which corresponds to the input dimension i + offset.
It is required that input_shape[i+offset]+paddings[i][0]+paddings[i][1] is divisible by block_size[i].
M must be 2 on Ascend.
Returns:
Tensor, the output tensor with the same data type as input.
Raises:
ValueError: If `block_size` is not one dimensional when `block_size` is a list or tuple.
ValueError: If the length of `block_size` is not 2 on Ascend.
ValueError: If the element of `block_size` is not an integer larger than 1.
ValueError: If shape of `paddings` is not (M, 2), where M is the length of `block_size`.
ValueError: If the element of `paddings` is not an integer larger than 0.
TypeError: If `block_size` is not one of list, tuple, int.
TypeError: If `paddings` is neither list nor tuple.
Supported Platforms:
``Ascend`` ``GPU`` ``CPU``
Examples:
>>> import numpy as np
>>> import mindspore
>>> from mindspore import Tensor, ops
>>> block_size = [2, 2]
>>> paddings = [[0, 0], [0, 0]]
>>> input_x = Tensor(np.array([[[[1, 2], [3, 4]]]]), mindspore.float32)
>>> output = ops.space_to_batch_nd(input_x, block_size, paddings)
>>> print(output)
[[[[1.]]]
[[[2.]]]
[[[3.]]]
[[[4.]]]]
"""
_space_to_batch_nd = _get_cache_prim(
P.SpaceToBatchND)(block_size, paddings)
return _space_to_batch_nd(input_x)
[docs]def batch_to_space_nd(input_x, block_shape, crops):
r"""
Divides batch dimension with blocks and interleaves these blocks back into spatial dimensions.
This operation will divide batch dimension N into blocks with block_shape, the output tensor's N dimension
is the corresponding number of blocks after division. The output tensor's :math:`w_1, ..., w_M` dimension is
the product of original :math:`w_1, ..., w_M` dimension and block_shape with given amount to crop from dimension,
respectively.
If the input shape is :math:`(n, c_1, ... c_k, w_1, ..., w_M)`, the output shape is
:math:`(n', c_1, ... c_k, w'_1, ..., w'_M)`, where
.. math::
\begin{array}{ll} \\
n' = n//(block\_shape[0]*...*block\_shape[M-1]) \\
w'_i = w_i*block\_shape[i-1]-crops[i-1][0]-crops[i-1][1]
\end{array}
Args:
input_x (Tensor): The input tensor. It must be greater or equal to 2-D tensor(equal to 4-D tensor on Ascend),
batch dimension must be divisible by product of `block_shape`.
block_shape (Union[list(int), tuple(int), int]): The block shape of dividing block with all value greater
than or equal to 1. If `block_shape` is a tuple or list, the length of `block_shape` is M corresponding
to the number of spatial dimensions. If `block_shape` is an int, the block size of M dimensions are the
same, equal to `block_shape`. In this case of Ascend, M must be 2.
crops (Union[list(int), tuple(int)]): The crops values for spatial dimensions, containing M subtraction list.
Each contains 2 integer values. All values must be >= 0. crops[i] specifies the crops values for spatial
dimension i, which corresponds to input dimension i + offset,where offset = N-M, and N is the number of
input dimensions. It is required that
:math:`input\_shape[i+offset]*block\_shape[i] > crops[i][0]+crops[i][1]`
Returns:
Tensor, the output tensor with the same type as input.
Raises:
TypeError: If `block_shape` is not one of list, tuple, int.
TypeError: If `crops` is neither list nor tuple.
ValueError: If `block_shape` is not one dimensional when `block_shape` is a list or tuple.
ValueError: If the length of `block_shape` is not 2 on Ascend.
ValueError: If the element of `block_shape` is not an integer larger than or euqal to 1.
ValueError: If shape of `crops` is not (M, 2), where M is the length of `block_shape`.
ValueError: If the element of `crops` is not an integer larger than or euqal to 0.
Supported Platforms:
``Ascend`` ``GPU`` ``CPU``
Examples:
>>> import mindspore
>>> import numpy as np
>>> from mindspore import Tensor, ops
>>> block_shape = [2, 2]
>>> crops = [[0, 0], [0, 0]]
>>> input_x = Tensor(np.array([[[[1]]], [[[2]]], [[[3]]], [[[4]]]]), mindspore.float32)
>>> output = ops.batch_to_space_nd(input_x, block_shape, crops)
>>> print(output)
[[[[1. 2.]
[3. 4.]]]]
"""
if isinstance(block_shape, Tensor):
return batch_to_space_nd_v2_(input_x, block_shape, crops)
_batch_to_space_nd = _get_cache_prim(P.BatchToSpaceND)(block_shape, crops)
return _batch_to_space_nd(input_x)
[docs]def matrix_diag(x, k=0, num_rows=-1, num_cols=-1, padding_value=0, align="RIGHT_LEFT"):
r"""
Returns a Tensor with the contents in `x` as k[0]-th to k[1]-th diagonals of a matrix, with everything else padded
with `padding_value`. `num_rows` and `num_cols` specify the dimension of the innermost matrix of the output. If both
are not specified, the op assumes the innermost matrix of output Tensor is square and infers its size from `k` and
the innermost dimension of `x`. If the `num_rows` and `num_cols` specify only one of them, the operator will derive
the smallest legal value as the dimension of output. Moreover, when only one diagonal is given
(k is an integer or k[0] == k[1]), the first to the second innermost dimension of `x` is the batch size. Otherwise,
the second innermost dimension is not a part of batch size.
Args:
x (Tensor): The diagonal Tensor.
k (Union[int, Tensor], optional): Diagonal offsets. A Tensor of type int32. Positive value means superdiagonal,
0 refers to the main diagonal, and negative value means subdiagonals. `k` can be a single integer
(for a single diagonal) or a pair of integers specifying the low and high ends of a matrix band.
k[0] must not be larger than k[1]. The value must be in the range of given or derivated `num_rows`
and `num_cols`, meaning value of k must be in (-num_rows, num_cols). Default: ``0`` .
num_rows (Union[int, Tensor], optional): The number of rows of the output Tensor. A Tensor of type int32 with
only one value. If `num_rows` is -1, indicating that the innermost matrix of the output Tensor is a square
matrix, and the real number of rows will be derivated by other inputs. That is
:math:`num\_rows = x.shape[-1] - min(k[1], 0)`. Otherwise, the value must be equal or greater than
:math:`x.shape[-1] - min(k[1], 0)`. Default: ``-1`` .
num_cols (Union[int, Tensor], optional): The number of columns of
the output Tensor. A Tensor of type int32 with only one value.
If `num_cols` is -1, indicating that the innermost matrix of the output
Tensor is a square matrix, and the real number of columns will be derivated by other inputs.
That is :math:`num\_cols = x.shape[-1] + max(k[0], 0)`. Otherwise, the value must be equal or
greater than :math:`x.shape[-1] - min(k[1], 0)`. Default: ``-1`` .
padding_value (Union[int, float, Tensor], optional): The number to fill the area outside the specified
diagonal band. A Tensor with only one value. Have the same dtype as x. Default: ``0`` .
align (str, optional): specifies how superdiagonals and subdiagonals should be aligned.
Supported values: ``"RIGHT_LEFT"`` , ``"LEFT_RIGHT"`` , ``"LEFT_LEFT"`` , ``"RIGHT_RIGHT"`` .
Default: ``"RIGHT_LEFT"`` .
- When set to "RIGHT_LEFT", the alignment of superdiagonals will be towards the right side
(padding the row on the left), while subdiagonals will be towards the left side
(padding the row on the right)
- When set to "LEFT_RIGHT", the alignment of superdiagonals will be towards the left side
(padding the row on the right), while subdiagonals will be towards the right side
(padding the row on the left)
- When set to "LEFT_LEFT", the alignment of both superdiagonals and subdiagonals will be towards
the left side(padding the row on the right).
- When set to "RIGHT_RIGHT", the alignment of both superdiagonals and subdiagonals will be towards
the right side(padding the row on the left).
Returns:
A Tensor. Has the same type as `x`.
Suppose `x` has r dimensions with shape :math:`(I, J, ..., M, N)` . The output Tensor has rank r + 1 with shape
:math:`(I, J, ..., M, num\_rows, num\_cols)` when only one diagonal is given (k is an integer or k[0] == k[1]).
Otherwise, it has rank r with shape :math:`(I, J, ..., num\_rows, num\_cols)` .
Raises:
TypeError: If `x` is not Tensor.
TypeError: If input `x` and `padding_value` are not the same dtype.
TypeError: If `k`, `num_rows` or `num_cols` is not int32 dtype.
ValueError: If rank of `k` is not equal to 0 or 1.
ValueError: If rank of `num_rows`, `num_cols` or `padding_value` is not equal to 0.
ValueError: If size of `k` is not equal to 1 or 2.
ValueError: If the value of `k` is not in (-num_rows, num_cols).
ValueError: If k[1] is not greater equal to k[0] when k[0] != k[1].
ValueError: If rank of `x` is not greater than or is equal to 1 when k is an integer or k[0] == k[1].
ValueError: If rank of `x` is not greater than or is equal to 2 when k[0] != k[1].
ValueError: If x.shape[-2] is not equal to k[1] - k[0] + 1 when k[0] != k[1].
ValueError: If `num_rows` and `num_cols` do not match the dimensions of `x` and the values of `k`.
ValueError: If `align` is not a string or not in the valid set of values.
Supported Platforms:
``Ascend`` ``GPU`` ``CPU``
Examples:
>>> import mindspore
>>> import numpy as np
>>> from mindspore import Tensor
>>> from mindspore import ops
>>> x = Tensor(np.array([[8, 9, 0],
... [1, 2, 3],
... [0, 4, 5]]), mindspore.float32)
>>> k =Tensor(np.array([-1, 1]), mindspore.int32)
>>> num_rows = Tensor(np.array(3), mindspore.int32)
>>> num_cols = Tensor(np.array(3), mindspore.int32)
>>> padding_value = Tensor(np.array(11), mindspore.float32)
>>> output = ops.matrix_diag(x, k, num_rows, num_cols, padding_value, align='LEFT_RIGHT')
>>> print(output)
[[ 1. 8. 11.]
[ 4. 2. 9.]
[11. 5. 3.]]
>>> print(output.shape)
(3, 3)
"""
if isinstance(k, int) and not isinstance(k, bool):
k = cast_(k, mstype.int32)
if isinstance(num_rows, int) and not isinstance(num_rows, bool):
num_rows = cast_(num_rows, mstype.int32)
if isinstance(num_cols, int) and not isinstance(num_cols, bool):
num_cols = cast_(num_cols, mstype.int32)
if isinstance(padding_value, (float, int)) and not isinstance(padding_value, bool):
padding_value = cast_(padding_value, x.dtype)
matrix_diag_v3 = _get_cache_prim(MatrixDiagV3)(align)
return matrix_diag_v3(x, k, num_rows, num_cols, padding_value)
[docs]def matrix_diag_part(x, k, padding_value, align="RIGHT_LEFT"):
r"""
Returns the diagonal part of input tensor.
Returns a tensor with the k[0]-th to k[1]-th diagonals of `x`. Some diagonals are shorter than
max_diag_len and need to be padded. Input k and padding_value must be const Tensor when taking Graph mode.
Args:
x (Tensor): The input Tensor with rank r, where r >= 2.
k (Tensor): A Tensor of type int32. Diagonal offset(s). Positive value means
superdiagonal, 0 refers to the main diagonal, and negative value means subdiagonals. k can be
a single integer (for a single diagonal) or a pair of integers specifying the low and high ends
of a matrix band. k[0] must not be larger than k[1]. The value of k has restructions, meaning
value of k must be in (-x.shape[-2], x.shape[-1]).
padding_value (Tensor): A Tensor with only one value. Have the same dtype as x.
The number to fill the area outside the specified diagonal band.
align (str, optional): An optional string from: ``"RIGHT_LEFT"`` , ``"LEFT_RIGHT"`` ,
``"LEFT_LEFT"`` , ``"RIGHT_RIGHT"`` . Align is a string specifying how superdiagonals and subdiagonals
should be aligned, respectively. ``"RIGHT_LEFT"`` aligns superdiagonals to the right (left-pads the row)
and subdiagonals to the left (right-pads the row). Default: ``"RIGHT_LEFT"`` . Default: ``"RIGHT_LEFT"``.
Returns:
A Tensor. Has the same type as `x`.
Assume `x` has r dimensions :math:`(I, J, ..., M, N)` . Let `max_diag_len` be the maximum length among all
diagonals to be extracted, :math:`max\_diag\_len = min(M + min(k[1], 0), N + min(-k[0], 0))`
Let `num_diags` be the number of diagonals to extract, :math:`num\_diags = k[1] - k[0] + 1`.
If :math:`num\_diags == 1`, the output tensor is of rank r - 1 with shape :math:`(I, J, ..., L, max\_diag\_len)`
Otherwise, the output tensor has rank r with dimensions :math:`(I, J, ..., L, num\_diags, max\_diag\_len)` .
Raises:
TypeError: If `x` is not Tensor.
TypeError: If input `x` and `padding_value` are not the same dtype.
TypeError: If `k` is not int32 dtype.
ValueError: If `align` is not a string or not in the valid range.
ValueError: If rank of `k` is not equal to 0 or 1.
ValueError: If rank of `padding_value` is not equal to 0.
ValueError: If rank of `x` is not greater equal to 2.
ValueError: If size of `k` is not equal to 1 or 2.
ValueError: If k[1] is not greater equal to k[0] in case the size of `k` is 2.
ValueError: If the value of `k` is not in (-x.shape[-2], x.shape[-1]).
Supported Platforms:
``Ascend`` ``GPU`` ``CPU``
Examples:
>>> import mindspore
>>> import numpy as np
>>> from mindspore import Tensor, ops
>>> x = Tensor(np.array([[1, 2, 3, 4],
... [5, 6, 7, 8],
... [9, 8, 7, 6]]), mindspore.float32)
>>> k =Tensor(np.array([1, 3]), mindspore.int32)
>>> padding_value = Tensor(np.array(9), mindspore.float32)
>>> output = ops.matrix_diag_part(x, k, padding_value, align='RIGHT_LEFT')
>>> print(output)
[[9. 9. 4.]
[9. 3. 8.]
[2. 7. 6.]]
>>> print(output.shape)
(3, 3)
"""
matrix_diag_part_v3 = _get_cache_prim(MatrixDiagPartV3)(align)
return matrix_diag_part_v3(x, k, padding_value)
[docs]def matrix_set_diag(x, diagonal, k=0, align="RIGHT_LEFT"): # pylint: disable=redefined-outer-name
r"""
Returns a batched matrix tensor with new batched diagonal values.
Given x and diagonal, this operation returns a tensor with the same shape and values as x, except for the specified
diagonals of the innermost matrices. These will be overwritten by the values in diagonal. Some diagonals are shorter
than max_diag_len and need to be padded.
The diagonal :math:`shape[-2]` must be equal to num_diags calculated by :math:`k[1] - k[0] + 1`.
The diagonal :math:`shape[-1]` must be
equal to the longest diagonal value max_diag_len calculated
by :math:`min(x.shape[-2] + min(k[1], 0), x.shape[-1] + min(-k[0], 0))`.
Let x have r + 1 dimensions :math:`(I, J, ..., L, M, N)` .
The diagonal tensor has rank r with shape :math:`(I, J, ..., L, max\_diag\_len)`
when k is an integer or :math:`k[0] == k[1]`. Otherwise, it has rank r + 1
with shape :math:`(I, J, ... L, num\_diags, max\_diag\_len)` .
Args:
x (Tensor): Rank r + 1, where r >= 1.
diagonal (Tensor): A Tensor. Have the same dtype as x. Rank r when k is an integer or :math:`k[0] == k[1]`.
Otherwise, it has rank r + 1.
k (Union[int, Tensor], optional): A int32 Scalar or int32 Tensor. Diagonal offset(s). Positive value means
superdiagonal, 0 refers to the main diagonal, and negative value means subdiagonals. k can be a
single integer (for a single diagonal) or a pair of integers specifying the low and high ends of
a matrix band. k[0] must not be larger than k[1].
The alue of k has restructions, meaning value of k must be in :math:`(-x.shape[-2], x.shape[-1])`.
Input k must be const Tensor when taking Graph mode. Default: ``0`` .
align (str, optional): An optional string from: ``"RIGHT_LEFT"`` (default), ``"LEFT_RIGHT"`` , ``"LEFT_LEFT"`` ,
``"RIGHT_RIGHT"`` . Align is a string specifying how superdiagonals and subdiagonals should be aligned,
respectively. ``"RIGHT_LEFT"`` aligns superdiagonals to the right (left-pads the row) and subdiagonals
to the left (right-pads the row).
Returns:
Tensor, The same type as x. Let x has r+1 dimensions :math:`(I, J, ..., L, M, N)` .
The output is a tensor of rank r+1 with dimensions :math:`(I, J, ..., L, M, N)` , the same as input x.
Raises:
TypeError: If input `x` or `diagonal` is not Tensor.
TypeError: If input `x` and `diagonal` are not the same dtype.
TypeError: If `k` is not int32 dtype.
ValueError: If `align` is not a string or not in the valid range.
ValueError: If rank of `k` is not equal to 0 or 1.
ValueError: If rank of `x` is not greater equal to 2.
ValueError: If size of `k` is not equal to 1 or 2.
ValueError: If k[1] is not greater equal to k[0] in case the size of `k` is 2.
ValueError: If the `diagonal` rank size don't match with input `x` rank size.
ValueError: If the `diagonal` shape value don't match with input `x` shape value.
ValueError: If the diagonal :math:`shape[-2]` is not equal to num_diags calculated by :math:`k[1]-k[0]+1`.
ValueError: If the value of `k` is not in :math:`(-x.shape[-2], x.shape[-1])`.
ValueError: If the diagonal.shape[-1] is not equal to the max_diag_len calculated by
:math:`min(x.shape[-2] + min(k[1],
0), x.shape[-1] + min(-k[0], 0))`.
Supported Platforms:
``Ascend`` ``GPU`` ``CPU``
Examples:
>>> import mindspore
>>> import numpy as np
>>> from mindspore import Tensor, ops
>>> x = Tensor(np.array([[7, 7, 7, 7],
... [7, 7, 7, 7],
... [7, 7, 7, 7]]), mindspore.float32)
>>> diagonal = Tensor(np.array([[0, 9, 1],
... [6, 5, 8],
... [1, 2, 3],
... [4, 5, 0]]), mindspore.float32)
>>> k = Tensor(np.array([-1, 2]), mindspore.int32)
>>> align = 'RIGHT_LEFT'
>>> output = ops.matrix_set_diag(x, diagonal, k, align)
>>> print(output)
[[1. 6. 9. 7.]
[4. 2. 5. 1.]
[7. 5. 3. 8.]]
>>> print(output.shape)
(3, 4)
"""
matrix_set_diag_v3_op = _get_cache_prim(MatrixSetDiagV3)(align)
if isinstance(k, int) and not isinstance(k, bool):
k = cast_(k, mstype.int32)
return matrix_set_diag_v3_op(x, diagonal, k)
[docs]def meshgrid(*inputs, indexing='xy'):
"""
Generates coordinate matrices from given coordinate tensors.
Given N one-dimensional coordinate tensors, returns a tuple outputs of N N-D
coordinate tensors for evaluating expressions on an N-D grid.
Args:
inputs (List[Tensor]): List of 1-D tensors.
The length of inputs should be greater than 1. The data type is Number.
Keyword Args:
indexing (str, optional): Cartesian ('xy', default) or
matrix ('ij') indexing of output. Valid options: xy' or ``'ij'``. In the 2-D case with
inputs of length `M` and `N`, the outputs are of shape :math:`(N, M)`
for ``'xy'`` indexing and :math:`(M, N)` for ``'ij'`` indexing. In the 3-D
case with inputs of length `M`, `N` and `P`, outputs are of shape
:math:`(N, M, P)` for ``'xy'`` indexing and :math:`(M, N, P)` for ``'ij'`` indexing.
Default: ``'xy'`` .
Returns:
Tensors, a Tuple of N N-D Tensor objects. The data type is the same with the Inputs.
Raises:
TypeError: If `indexing` is not a str or `inputs` is not a tuple.
ValueError: If `indexing` is neither ``'xy'`` nor ``'ij'``.
Supported Platforms:
``Ascend`` ``GPU`` ``CPU``
Examples:
>>> import numpy as np
>>> from mindspore import Tensor
>>> from mindspore import ops
>>> x = Tensor(np.array([1, 2, 3, 4]).astype(np.int32))
>>> y = Tensor(np.array([5, 6, 7]).astype(np.int32))
>>> z = Tensor(np.array([8, 9, 0, 1, 2]).astype(np.int32))
>>> output = ops.meshgrid(x, y, z, indexing='xy')
>>> print(output)
(Tensor(shape=[3, 4, 5], dtype=Int32, value=
[[[1, 1, 1, 1, 1],
[2, 2, 2, 2, 2],
[3, 3, 3, 3, 3],
[4, 4, 4, 4, 4]],
[[1, 1, 1, 1, 1],
[2, 2, 2, 2, 2],
[3, 3, 3, 3, 3],
[4, 4, 4, 4, 4]],
[[1, 1, 1, 1, 1],
[2, 2, 2, 2, 2],
[3, 3, 3, 3, 3],
[4, 4, 4, 4, 4]]]),
Tensor(shape=[3, 4, 5], dtype=Int32, value=
[[[5, 5, 5, 5, 5],
[5, 5, 5, 5, 5],
[5, 5, 5, 5, 5],
[5, 5, 5, 5, 5]],
[[6, 6, 6, 6, 6],
[6, 6, 6, 6, 6],
[6, 6, 6, 6, 6],
[6, 6, 6, 6, 6]],
[[7, 7, 7, 7, 7],
[7, 7, 7, 7, 7],
[7, 7, 7, 7, 7],
[7, 7, 7, 7, 7]]]),
Tensor(shape=[3, 4, 5], dtype=Int32, value=
[[[8, 9, 0, 1, 2],
[8, 9, 0, 1, 2],
[8, 9, 0, 1, 2],
[8, 9, 0, 1, 2]],
[[8, 9, 0, 1, 2],
[8, 9, 0, 1, 2],
[8, 9, 0, 1, 2],
[8, 9, 0, 1, 2]],
[[8, 9, 0, 1, 2],
[8, 9, 0, 1, 2],
[8, 9, 0, 1, 2],
[8, 9, 0, 1, 2]]]))
"""
meshgrid_op = _get_cache_prim(P.Meshgrid)(indexing)
return meshgrid_op(inputs)
[docs]def affine_grid(theta, size, align_corners=False):
r"""
Returns a 2D or 3D flow field (sampling grid) based on `theta`, a batch of affine matrices.
Args:
theta (Tensor): The input tensor of flow field whose dtype is float16, float32.
Input batch of affine matrices with shape :math:`(N, 2, 3)` for 2D grid or :math:`(N, 3, 4)` for 3D grid.
size (tuple[int]): The target output image size.
The value of target output with format :math:`(N, C, H, W)` for 2D grid or :math:`(N, C, D, H, W)` for 3D
grid.
align_corners (bool, optional): Geometrically, each pixel of input is viewed as a squqre instead of dot.
If ``True`` , consider extremum -1 and 1 referring to the centers of the pixels rather than pixel corners.
The default value is ``False`` , extremum -1 and 1 refer to the corners of the pixels, so that sampling is
irrelevant to resolution of the image. Default: ``False`` .
Returns:
Tensor, a tensor whose data type is same as 'theta', and the shape is :math:`(N, H, W, 2)` for 2D grid
or :math:`(N, D, H, W, 3)` for 3D grid.
Raises:
TypeError: If `theta` is not a Tensor or `size` is not a tuple.
ValueError: If the shape of `theta` is not :math:`(N, 2, 3)` or :math:`(N, 3, 4)`.
ValueError: If the size of `size` is not 4 or 5.
ValueError: If the shape of `theta` is :math:`(N, 2, 3)`, the size of `size` is not 4;
If the shape of `theta` is :math:`(N, 3, 4)`, the size of `size` is not 5.
ValueError: If the size[0] is not equal to the shape[0] of theta.
Supported Platforms:
``Ascend`` ``GPU`` ``CPU``
Examples:
>>> import mindspore
>>> from mindspore import Tensor
>>> from mindspore import ops
>>> theta = Tensor([[[0.8, 0.5, 0],[-0.5, 0.8, 0]]], mindspore.float32)
>>> out_size = (1, 3, 2, 3)
>>> output = ops.affine_grid(theta, out_size, False)
>>> print(output)
[[[[-0.78333336 -0.06666666]
[-0.25 -0.4 ]
[ 0.28333336 -0.73333335]]
[[-0.28333336 0.73333335]
[ 0.25 0.4 ]
[ 0.78333336 0.06666666]]]]
"""
affine_grid_op = AffineGrid(align_corners)
return affine_grid_op(theta, size)
[docs]def unsorted_segment_min(x, segment_ids, num_segments):
r"""
Computes the minimum of a tensor along segments.
The following figure shows the calculation process of unsorted_segment_min:
.. image:: UnsortedSegmentMin.png
.. math::
\text { output }_i=\text{min}_{j \ldots} \text { data }[j \ldots]
where :math:`min` over tuples :math:`j...` such that :math:`segment\_ids[j...] == i`.
Note:
- If the segment_id i is absent in the segment_ids, then output[i] will be filled with
the maximum value of the x's type.
- The `segment_ids` must be non-negative tensor.
Args:
x (Tensor): The shape is :math:`(x_1, x_2, ..., x_R)`. With float16, float32 or int32 data type.
segment_ids (Tensor): TThe label indicates the segment to which each element belongs.
Set the shape as :math:`(x_1, x_2, ..., x_N)`, where 0 < N <= R.
num_segments (Union[int, Tensor], optional): Set :math:`z` as num_segments, it can be an int or 0-D Tensor.
Returns:
Tensor, the shape is :math:`(z, x_{N+1}, ..., x_R)`.
Raises:
TypeError: If `num_segments` is not an int.
Supported Platforms:
``Ascend`` ``GPU`` ``CPU``
Examples:
>>> from mindspore import Tensor
>>> from mindspore import ops
>>> import numpy as np
>>> x = Tensor(np.array([[1, 2, 3], [4, 5, 6], [4, 2, 1]]).astype(np.float32))
>>> segment_ids = Tensor(np.array([0, 1, 1]).astype(np.int32))
>>> num_segments = 2
>>> output = ops.unsorted_segment_min(x, segment_ids, num_segments)
>>> print(output)
[[1. 2. 3.]
[4. 2. 1.]]
"""
return unsorted_segment_min_(x, segment_ids, num_segments)
[docs]def unsorted_segment_max(x, segment_ids, num_segments):
r"""
Computes the maximum along segments of a tensor.
The following figure shows the calculation process of unsorted_segment_max:
.. image:: UnsortedSegmentMax.png
.. math::
\text { output }_i=\text{max}_{j \ldots} \text { data }[j \ldots]
where :math:`max` over tuples :math:`j...` such that :math:`segment\_ids[j...] == i`.
Note:
- If the segment_id i is absent in the segment_ids, then output[i] will be filled with
the minimum value of the x's type.
- The `segment_ids` must be non-negative tensor.
Args:
x (Tensor): The shape is :math:`(x_1, x_2, ..., x_R)`. With float16, float32 or int32 data type.
segment_ids (Tensor): TThe label indicates the segment to which each element belongs.
Set the shape as :math:`(x_1, x_2, ..., x_N)`, where 0 < N <= R.
num_segments (Union[int, Tensor], optional): Set :math:`z` as num_segments, it can be an int or 0-D Tensor.
Returns:
Tensor, the shape is :math:`(z, x_{N+1}, ..., x_R)`.
Raises:
TypeError: If `num_segments` is not an int.
Supported Platforms:
``Ascend`` ``GPU`` ``CPU``
Examples:
>>> from mindspore import Tensor
>>> from mindspore import ops
>>> import numpy as np
>>> x = Tensor(np.array([[1, 2, 3], [4, 5, 6], [4, 2, 1]]).astype(np.float32))
>>> segment_ids = Tensor(np.array([0, 1, 1]).astype(np.int32))
>>> num_segments = 2
>>> output = ops.unsorted_segment_max(x, segment_ids, num_segments)
>>> print(output)
[[1. 2. 3.]
[4. 5. 6.]]
"""
return unsorted_segment_max_(x, segment_ids, num_segments)
[docs]def unsorted_segment_prod(x, segment_ids, num_segments):
r"""
Computes the product of a tensor along segments.
The following figure shows the calculation process of unsorted_segment_prod:
.. image:: UnsortedSegmentProd.png
Note:
- If the segment_id i is absent in the segment_ids, then output[i] will be filled with 1.
- The `segment_ids` must be non-negative tensor.
Args:
x (Tensor): The shape is :math:`(x_1, x_2, ..., x_R)`. With float16, float32 or int32 data type.
segment_ids (Tensor): TThe label indicates the segment to which each element belongs.
Set the shape as :math:`(x_1, x_2, ..., x_N)`, where 0 < N <= R. The data type must be int32.
num_segments (Union[int, Tensor], optional): Set :math:`z` as num_segments, it can be an int or 0-D Tensor.
Returns:
Tensor, the shape is :math:`(z, x_{N+1}, ..., x_R)`.
Raises:
TypeError: If `num_segments` is not an int.
Supported Platforms:
``Ascend`` ``GPU`` ``CPU``
Examples:
>>> from mindspore import Tensor
>>> from mindspore import ops
>>> import numpy as np
>>> x = Tensor(np.array([[1, 2, 3], [4, 5, 6], [4, 2, 1]]).astype(np.float32))
>>> segment_ids = Tensor(np.array([0, 1, 0]).astype(np.int32))
>>> num_segments = 2
>>> output = ops.unsorted_segment_prod(x, segment_ids, num_segments)
>>> print(output)
[[4. 4. 3.]
[4. 5. 6.]]
"""
return unsorted_segment_prod_(x, segment_ids, num_segments)
[docs]def index_fill(x, axis, index, value):
"""
Fills the elements under the `axis` dimension of the input Tensor `x` with the input `value`
by selecting the indices in the order given in `index`.
Args:
x (Tensor): Input Tensor. The supported data type is Number or Bool.
axis (Union[int, Tensor]): Dimension along which to fill the input Tensor. Only supports
an int number or a 0-dimensional Tensor, whose data type is int32 or int64.
index (Tensor): Indices of the input Tensor to fill in. The dtype must be int32.
value (Union[bool, int, float, Tensor]): Value to fill the returned Tensor. If `value` is
a Tensor, it must be a 0-dimensional Tensor and has the same dtype as `x`. Otherwise,
the `value` will be cast to a 0-dimensional Tensor with the same data type as `x`.
Returns:
Tensor, has the same dtype and shape as input Tensor.
Raises:
TypeError: If `x` is not a Tensor.
TypeError: If `axis` is neither int number nor Tensor.
TypeError: When `axis` is a Tensor, its dtype is not int32 or int64.
TypeError: If `index` is not a Tensor.
TypeError: If dtype of `index` is not int32.
TypeError: If `value` is not a bool, int, float, or Tensor.
TypeError: When `value` is a Tensor, the dtype of `x` and `value` are not the same.
ValueError: If `axis` is a Tensor and its rank is not equal to 0.
ValueError: If the rank of `index` is greater than 1D.
ValueError: When `value` is a Tensor and its rank is not equal to 0.
RuntimeError: If the value of `axis` is out the range of `[-x.ndim, x.ndim - 1]`.
RuntimeError: If the values of `index` are out the range of `[-x.shape[axis], x.shape[axis]-1]`.
Supported Platforms:
``Ascend`` ``GPU`` ``CPU``
Examples:
>>> import mindspore
>>> import numpy as np
>>> from mindspore import ops
>>> from mindspore import Tensor
>>> x = Tensor(np.array([[1, 2, 3], [4, 5, 6], [7, 8, 9]]).astype(np.float32))
>>> index = Tensor([0, 2], mindspore.int32)
>>> value = Tensor(-2.0, mindspore.float32)
>>> y = ops.index_fill(x, 1, index, value)
>>> print(y)
[[-2. 2. -2.]
[-2. 5. -2.]
[-2. 8. -2.]]
"""
if isinstance(axis, int) and not isinstance(axis, bool):
axis = cast_(axis, mstype.int32)
if isinstance(value, (bool, float, int)):
value = cast_(value, x.dtype)
return index_fill_(x, axis, index, value)
@constexpr
def _check_check_axis_in_range(axis, ndim):
"""Checks axes are with the bounds of ndim"""
axis = validator.check_axis_in_range(axis, ndim)
return axis
[docs]def index_select(input, axis, index):
"""
Generates a new Tensor that accesses the values of `input` along the specified `axis` dimension
using the indices specified in `index`. The new Tensor has the same number of dimensions as `input`,
with the size of the `axis` dimension being equal to the length of `index`, and the size of all other
dimensions will be unchanged from the original `input` Tensor.
.. note::
The value of index must be in the range of `[0, input.shape[axis])`, the result is undefined out of range.
Args:
input (Tensor): The input Tensor.
axis (int): The dimension to be indexed.
index (Tensor): A 1-D Tensor with the indices to access in `input` along the specified axis.
Returns:
Tensor, has the same dtype as input Tensor.
Raises:
TypeError: If `input` or `index` is not a Tensor.
TypeError: If `axis` is not int number.
ValueError: If the value of `axis` is out the range of `[-input.ndim, input.ndim - 1]`.
ValueError: If the dimension of `index` is not equal to 1.
Supported Platforms:
``Ascend`` ``GPU`` ``CPU``
Examples:
>>> import mindspore
>>> from mindspore import Tensor, ops
>>> import numpy as np
>>> input = Tensor(np.arange(16).astype(np.float32).reshape(2, 2, 4))
>>> print(input)
[[[ 0. 1. 2. 3.]
[ 4. 5. 6. 7.]]
[[ 8. 9. 10. 11.]
[12. 13. 14. 15.]]]
>>> index = Tensor([0,], mindspore.int32)
>>> y = ops.index_select(input, 1, index)
>>> print(y)
[[[ 0. 1. 2. 3.]]
[[ 8. 9. 10. 11.]]]
"""
if not (isinstance(input, Tensor) and isinstance(index, Tensor)):
raise TypeError(
f"For 'index_select', `input` and `index` must be all tensors.")
if index.ndim != 1:
raise ValueError(
f"For 'index_select', the dimension of `index` must be 1, but got {index.ndim}")
axis = _check_check_axis_in_range(axis, input.ndim)
return gather_(input, index, axis)
[docs]def population_count(input_x):
r"""
Computes element-wise population count(a.k.a bitsum, bitcount).
For each entry in `input_x`, calculates the number of 1 bits in the binary representation of that entry.
Args:
input_x (Tensor): Tensor of any dimension. The data type must be int16 or uint16 (Ascend).
The data type must be int8, int16, int32, int64, uint8, uint16, uint32, uint64 (CPU and GPU).
Returns:
Tensor, with the same shape as the input, and the data type is uint8.
Raises:
TypeError: If `input_x` is not a Tensor.
TypeError: If dtype of `input_x` is not int16, uint16 (Ascend).
TypeError: If dtype of `input_x` is not int8, int16, int32, int64, uint8, uint16, uint32, uint64 (CPU and GPU).
Supported Platforms:
``Ascend`` ``GPU`` ``CPU``
Examples:
>>> import mindspore
>>> from mindspore import Tensor, ops
>>> input_x = Tensor([0, 1, 3], mindspore.int16)
>>> output = ops.population_count(input_x)
>>> print(output)
[0 1 2]
"""
return population_count_(input_x)
##############################
# Type Conversion Functions.
##############################
[docs]def is_tensor(obj):
r"""
Check whether the input object is a :class:`mindspore.Tensor` .
Args:
obj (Object): input object.
Returns:
Bool. Return True if `obj` is a Tensor, otherwise, return False.
Supported Platforms:
``Ascend`` ``GPU`` ``CPU``
Examples:
>>> from mindspore import Tensor, ops
>>> a = Tensor([1.9, 2.2, 3.1])
>>> ops.is_tensor(a)
True
"""
return isinstance(obj, Tensor)
[docs]def is_nonzero(input):
"""
Determine whether the input Tensor contains 0 or False. The input can only be a single element.
Args:
input (Tensor): The input tensor.
Returns:
Bool, returns False if the input Tensor contains a unit element of 0 or a single element of False,
otherwise returns True.
Raises:
TypeError: If `input` is not Tensor.
ValueError: If the element number of `input` is not equal to 1.
Supported Platforms:
``Ascend`` ``GPU`` ``CPU``
Examples:
>>> from mindspore import Tensor, ops
>>> x1 = Tensor([[[False]]])
>>> x2 = Tensor([[3.5]])
>>> out1 = ops.is_nonzero(x1)
>>> print(out1)
False
>>> out2 = ops.is_nonzero(x2)
>>> print(out2)
True
"""
if not isinstance(input, Tensor):
raise TypeError(
f'For is_nonzero, the input must be a Tensor, but got {type(input)}.')
if input.numel() != 1:
raise ValueError(
f"For is_nonzero, the numel of input must be 1, but got {input.numel()}.")
out = ops.squeeze(input)
return bool(out)
[docs]def tensor_scatter_mul(input_x, indices, updates):
r"""
Creates a new tensor by multiplying the values from the positions in `input_x` indicated by
`indices`, with values from `updates`. When divided values are provided for the same
index, the result of the update will multiply these values respectively. Except that
the updates are applied on output `Tensor` instead of input `Parameter`.
The last axis of `indices` is the depth of each index vectors. For each index vector,
there must be a corresponding value in `updates`. The shape of `updates` should be
equal to the shape of `input_x[indices]`. For more details, see Examples.
.. math::
output\left [indices \right ] = input\_x\times update
Note:
- If some values of the `indices` are out of bound, instead of raising an index error,
the corresponding `updates` will not be updated to `input_x`.
Args:
input_x (Tensor): The input tensor. The dimension of `input_x` must be no less than indices.shape[-1].
indices (Tensor): The index of input tensor whose data type is int32 or int64. The rank must be at least 2.
updates (Tensor): The tensor to update the input tensor, has the same type as `input_x`,
and the shape of `updates` should be equal to
:math:`indices.shape[:-1] + input\_x.shape[indices.shape[-1]:]`.
Returns:
Tensor, has the same shape and type as `input_x`.
Raises:
TypeError: If dtype of `indices` is neither int32 nor int64.
ValueError: If length of shape of `input_x` is less than the last dimension of shape of `indices`.
RuntimeError: If a value of `indices` is not in `input_x` on CPU backend.
Supported Platforms:
``GPU`` ``CPU``
Examples:
>>> import mindspore
>>> import numpy as np
>>> from mindspore import Tensor, ops
>>> input_x = Tensor(np.array([[-0.1, 0.3, 3.6], [0.4, 0.5, -3.2]]), mindspore.float32)
>>> indices = Tensor(np.array([[0, 0], [0, 0]]), mindspore.int32)
>>> updates = Tensor(np.array([1.0, 2.2]), mindspore.float32)
>>> # Next, demonstrate the approximate operation process of this operator:
>>> # 1, indices[0] = [0, 0], indices[1] = [0, 0]
>>> # 2, And input_x[0, 0] = -0.1
>>> # 3, So input_x[indices] = [-0.1, -0.1]
>>> # 4, Satisfy the above formula: input_x[indices].shape=(2) == updates.shape=(2)
>>> # 5, Perform the multiply operation for the first time:
>>> # first_input_x = input_x[0][0] * updates[0] = [[-0.1, 0.3, 3.6], [0.4, 0.5, -3.2]]
>>> # 6, Perform the multiply operation for the second time:
>>> # second_input_x = input_x[0][0] * updates[1] = [[-0.22, 0.3, 3.6], [0.4, 0.5, -3.2]]
>>> output = ops.tensor_scatter_mul(input_x, indices, updates)
>>> print(output)
[[-0.22 0.3 3.6 ]
[ 0.4 0.5 -3.2 ]]
"""
return tensor_scatter_mul_(input_x, indices, updates)
[docs]def tensor_scatter_div(input_x, indices, updates):
r"""
Creates a new tensor by dividing the values from the positions in `input_x` indicated by
`indices`, with values from `updates`. When divided values are provided for the same
index, the result of the update will be to divided these values respectively. Except that
the updates are applied on output `Tensor` instead of input `Parameter`.
The last axis of `indices` is the depth of each index vectors. For each index vector,
there must be a corresponding value in `updates`. The shape of `updates` should be
equal to the shape of `input_x[indices]`. For more details, see Examples.
.. math::
output\left [indices \right ] = input\_x \div update
Note:
- On GPU, if some values of the `indices` are out of bound, instead of raising an index error,
the corresponding `updates` will not be updated to self tensor.
- On CPU, if some values of the `indices` are out of bound, raising an index error.
- On Ascend, out of bound checking is not supported, if some values of the `indices` are out of bound,
unknown errors may be caused.
- The operator can't handle division by 0 exceptions, so the user needs to make sure
there is no 0 value in `updates`.
Args:
input_x (Tensor): The input tensor. The dimension of input_x must be no less than indices.shape[-1].
indices (Tensor): The index of input tensor whose data type is int32 or int64.
The rank must be at least 2.
updates (Tensor): The tensor to update the `input_x` tensor, has the same type as `input_x`.
And the shape of `updates` should be
equal to :math:`indices.shape[:-1] + input\_x.shape[indices.shape[-1]:]`.
Returns:
Tensor, has the same shape and type as `input_x`.
Raises:
TypeError: If dtype of `indices` is neither int32 nor int64.
ValueError: If length of shape of `input_x` is less than the last dimension of shape of `indices`.
RuntimeError: If a value of `indices` is not in `input_x`.
Supported Platforms:
``GPU`` ``CPU``
Examples:
>>> import numpy as np
>>> import mindspore
>>> from mindspore import Tensor, nn, ops
>>> input_x = Tensor(np.array([[-0.1, 0.3, 3.6], [0.4, 0.5, -3.2]]), mindspore.float32)
>>> indices = Tensor(np.array([[0, 0], [0, 0]]), mindspore.int32)
>>> updates = Tensor(np.array([1.0, 2.0]), mindspore.float32)
>>> output = ops.tensor_scatter_div(input_x, indices, updates)
>>> print(output)
[[-0.05 0.3 3.6 ]
[ 0.4 0.5 -3.2 ]]
"""
return tensor_scatter_div_(input_x, indices, updates)
def scalar_to_array(input_x):
"""
The interface is deprecated. Please use the :func:`mindspore.ops.scalar_to_tensor` instead.
"""
return P.ScalarToArray()(input_x)
[docs]def scalar_to_tensor(input_x, dtype=mstype.float32):
"""
Converts a scalar to a `Tensor`, and converts the data type to the specified type.
Args:
input_x (Union[bool, int, float]): The input is a scalar. Only constant value is allowed.
dtype (mindspore.dtype): The target data type. Only constant value is allowed. Default: ``mstype.float32``.
Returns:
Tensor. 0-D Tensor and the content is the input.
Raises:
TypeError: If `input_x` is neither bool nor int nor float.
Supported Platforms:
``Ascend`` ``GPU`` ``CPU``
Examples:
>>> import mindspore
>>> from mindspore import ops
>>> data = 1
>>> output = ops.scalar_to_tensor(data, mindspore.float32)
>>> print(output)
1.0
"""
return scalar_to_tensor_(input_x, dtype)
[docs]def tuple_to_array(input_x):
"""
Converts a tuple to a tensor.
If the type of the first number in the tuple is integer, the data type of the output tensor is int.
Otherwise, the data type of the output tensor is float.
Args:
input_x (tuple): A tuple of numbers. These numbers have the same type. Only constant value is allowed.
The shape is :math:`(N,*)` where :math:`*` means any number of additional dimensions.
Returns:
Tensor, if the input tuple contains `N` numbers, then the shape of the output tensor is (N,).
Raises:
TypeError: If `input_x` is not a tuple.
ValueError: If length of `input_x` is less than or equal to 0.
Supported Platforms:
``Ascend`` ``GPU`` ``CPU``
Examples:
>>> input_x = (1,2,3)
>>> print(type(input_x))
<class 'tuple'>
>>> output = ops.tuple_to_array(input_x)
>>> print(type(output))
<class 'mindspore.common.tensor.Tensor'>
>>> print(output)
[1 2 3]
"""
if isinstance(input_x[0], int):
dtype = mstype.int32
else:
dtype = mstype.float32
return tuple_to_tensor_(input_x, dtype)
[docs]def diagflat(input, offset=0):
r"""
Create a 2-D Tensor which diagonal is the flattened `input` .
Args:
input (Tensor): Input Tensor, which is flattened and set as the diagonal of the output.
offset (int, optional): `offset` controls which diagonal to choose. Default: ``0`` .
- When `offset` is zero, the diagonal chosen is the main diagonal.
- When `offset` is a positive integer, the diagonal chosen is up the main diagonal.
- When `offset` is a negative integer, the diagonal chosen is down the main diagonal.
Returns:
The 2-D Tensor, whose diagonal is the flattened `input`.
Raises:
TypeError: If `input` is not a tensor.
TypeError: If `offset` is not an integer.
Supported Platforms:
``Ascend`` ``GPU`` ``CPU``
Examples:
>>> import mindspore
>>> from mindspore import Tensor, ops
>>> x = Tensor([1, 2], mindspore.float32)
>>> output = ops.diagflat(x, 1)
>>> print(output)
[[0. 1. 0.]
[0. 0. 2.]
[0. 0. 0.]]
"""
if not isinstance(input, Tensor):
raise TypeError(
f"For diagflat, the input x must be tensor, but got {type(input)}")
if not isinstance(offset, int):
raise TypeError(
f"For diagflat, the offset must be int, but got {type(offset)}")
offset_abs = abs(offset)
if input.size == 0:
return zeros((offset_abs, offset_abs), input.dtype)
input = input.ravel()
res = diag(input)
if offset != 0:
pad_y = zeros((input.size + offset_abs, offset_abs), input.dtype)
pad_x = zeros((offset_abs, input.size), input.dtype)
if offset < 0:
res = cat((pad_x, res), axis=0)
res = cat((res, pad_y), axis=1)
else:
res = cat((res, pad_x), axis=0)
res = cat((pad_y, res), axis=1)
return res
[docs]def col2im(input_x, output_size, kernel_size, dilation, padding_value, stride):
"""
Combines an array of sliding local blocks into a large containing tensor.
Args:
input_x (Tensor): 4D tensor with data type float16 or float32.
output_size (Tensor): 1D tensor with 2 elements of data type int.
kernel_size (Union[int, tuple[int], list[int]]): The size of the kernel, should be two int
for height and width. If type is int, it means that height equal with width. Must be specified.
dilation (Union[int, tuple[int], list[int]]): The size of the dilation, should be two int
for height and width. If type is int, it means that height equal with width.
padding_value (Union[int, tuple[int], list[int]]): The size of the padding, should be two int
for height and width. If type is int, it means that height equal with width.
stride (Union[int, tuple[int], list[int]]): The size of the stride, should be two int
for height and width. If type is int, it means that height equal with width.
Returns:
A 4D Tensor, with same type as 'input_x'.
Raises:
TypeError: If :attr:`kernel_size`, `dilation`, `padding_value`, `stride` data type is not in
Union[int, tuple[int], list[int]].
ValueError: If :attr:`kernel_size`, `dilation`, `padding_value`, `stride` value is not
greater than zero or elements number more than 2.
ValueError: If :attr:`padding_value` value is less than zero or elements number more than 2.
ValueError: If input_x.shape[2] != kernel_size[0] * kernel_size[1].
ValueError: If input_x.shape[3] does not match the calculated number of sliding blocks.
Supported Platforms:
``Ascend`` ``GPU`` ``CPU``
Examples:
>>> import numpy as np
>>> from mindspore import Tensor, ops
>>> from mindspore import dtype as mstype
>>> x = Tensor(input_data=np.random.rand(16, 16, 4, 25), dtype=mstype.float32)
>>> output_size = Tensor(input_data=[8, 8], dtype=mstype.int32)
>>> output = ops.col2im(x, output_size, [2, 2], [2, 2], [2, 2], [2, 2])
>>> print(output.shape)
(16, 16, 8, 8)
"""
c2i = _get_cache_prim(Col2Im)(kernel_size, dilation, padding_value, stride)
return c2i(input_x, output_size)
def _split_int(x, split_size_or_sections, axis):
"""
Splits the input tensor `x` into multiple sub-tensors along the axis according to the given `split_size_or_sections`
with int type.
"""
arr_shape = x.shape
length_along_dim = arr_shape[axis]
if length_along_dim == 0:
res = _get_cache_prim(P.Split)(axis)(x)
elif split_size_or_sections > length_along_dim:
res = _get_cache_prim(P.Split)(axis, 1)(x)
elif length_along_dim % split_size_or_sections == 0:
sections = length_along_dim // split_size_or_sections
res = _get_cache_prim(P.Split)(axis, sections)(x)
else:
num_sections = length_along_dim // split_size_or_sections
length1 = num_sections * split_size_or_sections
length2 = length_along_dim - length1
start1 = _list_comprehensions(rank_(x), 0, True)
size1 = _tuple_setitem(arr_shape, axis, length1)
start2 = _tuple_setitem(start1, axis, length1)
size2 = _tuple_setitem(arr_shape, axis, length2)
res = _get_cache_prim(P.Split)(axis, num_sections)(tensor_slice(x, start1, size1)) + \
_get_cache_prim(P.Split)(axis, 1)(tensor_slice(x, start2, size2))
return res
def _split_sub_tensors(x, split_size_or_sections, axis):
"""
Splits the input tensor `x` into multiple sub-tensors along the axis according to the given `split_size_or_sections`
with type of tuple or list.
"""
new_indices = [0]
for i, split_size in enumerate(split_size_or_sections):
new_indices.append(new_indices[i] + split_size)
new_indices = new_indices[1:]
sub_tensors = []
strides = _list_comprehensions(x.ndim, 1, True)
begin = _list_comprehensions(x.ndim, 0)
end = _list_comprehensions(x.shape)
for i in ms_arrange(len(new_indices)):
idx = new_indices[i]
begin[axis] = 0 if i == 0 else new_indices[i - 1]
end[axis] = idx
sliced_tensor = strided_slice(x, tuple(begin), tuple(end), strides)
sub_tensors.append(sliced_tensor)
return sub_tensors
[docs]def split(tensor, split_size_or_sections, axis=0):
"""
Splits the Tensor into chunks along the given axis.
Args:
tensor (Tensor): A Tensor to be divided.
split_size_or_sections (Union[int, tuple(int), list(int)]):
If `split_size_or_sections` is an int type, `tensor` will be split into equally sized chunks,
each chunk with size `split_size_or_sections`. Last chunk will be smaller than `split_size_or_sections`
if `tensor.shape[axis]` is not divisible by `split_size_or_sections`.
If `split_size_or_sections` is a list type, then `tensor` will be split into len(split_size_or_sections)
chunks with sizes `split_size_or_sections` along the given `axis`.
axis (int): The axis along which to split. Default: ``0`` .
Returns:
A tuple of sub-tensors.
Raises:
TypeError: If argument `tensor` is not Tensor.
TypeError: If argument `axis` is not Tensor.
ValueError: If argument `axis` is out of range of :math:`[-tensor.ndim, tensor.ndim)` .
TypeError: If each element in `split_size_or_sections` is not integer.
TypeError: If argument `split_size_or_sections` is not int, tuple(int) or list(int).
ValueError: The sum of `split_size_or_sections` is not equal to x.shape[axis].
Supported Platforms:
``Ascend`` ``GPU`` ``CPU``
Examples:
>>> import numpy as np
>>> from mindspore import ops, Tensor
>>> input_x = np.arange(9).astype("float32")
>>> output = ops.split(Tensor(input_x), 3)
>>> print(output)
(Tensor(shape=[3], dtype=Float32, value= [ 0.00000000e+00, 1.00000000e+00, 2.00000000e+00]),
Tensor(shape=[3], dtype=Float32, value= [ 3.00000000e+00, 4.00000000e+00, 5.00000000e+00]),
Tensor(shape=[3], dtype=Float32, value= [ 6.00000000e+00, 7.00000000e+00, 8.00000000e+00]))
"""
if not isinstance(tensor, Tensor):
raise TypeError(f'expect `tensor` is a Tensor, but got {type(tensor)}')
if type(axis) is not int:
raise TypeError(
f"Type of Argument `axis` should be integer but got {type(axis)}")
arr_axis = _canonicalize_axis(axis, tensor.ndim)
if type(split_size_or_sections) is int:
if split_size_or_sections > 0:
res = _split_int(tensor, split_size_or_sections, arr_axis)
else:
raise ValueError(f"For split, the value of 'split_size_or_sections' must be more than zero, "
f"but got {split_size_or_sections}.")
elif isinstance(split_size_or_sections, (list, tuple)):
for item in split_size_or_sections:
if type(item) is not int:
raise TypeError(
f"Each element in 'split_size_or_sections' should be integer, but got {type(item)}.")
if item < 0:
raise TypeError(f"Each element in 'split_size_or_sections' should be non-negative, "
f"but got {split_size_or_sections}.")
if sum(split_size_or_sections) != tensor.shape[arr_axis]:
raise ValueError(f"The sum of 'split_size_or_sections' should be equal to {tensor.shape[arr_axis]}, "
f"but got {sum(split_size_or_sections)}.")
res = _split_sub_tensors(tensor, split_size_or_sections, arr_axis)
else:
raise TypeError(f"Type of Argument `split_size_or_sections` should be integer, tuple(int) or list(int), "
f"but got {type(split_size_or_sections)}")
return tuple(res)
def split_ext(tensor, split_size_or_sections, axis=0):
"""
Splits the Tensor into chunks along the given axis.
Args:
tensor (Tensor): A Tensor to be divided.
split_size_or_sections (Union[int, tuple(int), list(int)]):
If `split_size_or_sections` is an int type, `tensor` will be split into equally sized chunks,
each chunk with size `split_size_or_sections`. Last chunk will be smaller than `split_size_or_sections`
if `tensor.shape[axis]` is not divisible by `split_size_or_sections`.
If `split_size_or_sections` is a list type, then `tensor` will be split into len(split_size_or_sections)
chunks with sizes `split_size_or_sections` along the given `axis`.
axis (int): The axis along which to split. Default: ``0`` .
Returns:
A tuple of sub-tensors.
Raises:
TypeError: If argument `tensor` is not Tensor.
TypeError: If argument `axis` is not int.
ValueError: If argument `axis` is out of range of :[-tensor.ndim, tensor.ndim).
TypeError: If each element in `split_size_or_sections` is not integer.
TypeError: If argument `split_size_or_sections` is not int, tuple(int) or list(int).
ValueError: The sum of `split_size_or_sections` is not equal to x.shape[axis].
Supported Platforms:
``Ascend``
Examples:
>>> import numpy as np
>>> from mindspore import ops, Tensor
>>> input_x = np.arange(9).astype("float32")
>>> output = ops.split_ext(Tensor(input_x), 3)
>>> print(output)
(Tensor(shape=[3], dtype=Float32, value= [ 0.00000000e+00, 1.00000000e+00, 2.00000000e+00]),
Tensor(shape=[3], dtype=Float32, value= [ 3.00000000e+00, 4.00000000e+00, 5.00000000e+00]),
Tensor(shape=[3], dtype=Float32, value= [ 6.00000000e+00, 7.00000000e+00, 8.00000000e+00]))
"""
if isinstance(split_size_or_sections, int):
res = split_tensor(tensor, split_size_or_sections, axis)
elif isinstance(split_size_or_sections, (list, tuple)):
res = split_with_size(tensor, split_size_or_sections, axis)
else:
raise TypeError(f"Type of Argument `split_size_or_sections` should be integer, tuple(int) or list(int), "
f"but got {type(split_size_or_sections)}")
return res
[docs]def tril(input, diagonal=0): # pylint: disable=redefined-outer-name
"""
Returns the lower triangle part of `input` (elements that contain the diagonal and below),
and set the other elements to zeros.
Args:
input (Tensor): A Tensor with shape :math:`(x_1, x_2, ..., x_R)`. The rank must be at least 2.
Supporting all number types including bool.
diagonal (int, optional): An optional attribute indicates the diagonal to consider, default: 0,
indicating the main diagonal.
Returns:
Tensor, the same shape and data type as the `input`.
Raises:
TypeError: If `input` is not a Tensor.
TypeError: If `diagonal` is not an int.
TypeError: If the type of `input` is neither number nor bool.
ValueError: If the rank of `input` is less than 2.
Supported Platforms:
``Ascend`` ``GPU`` ``CPU``
Examples:
>>> import numpy as np
>>> from mindspore import Tensor, ops
>>> x = Tensor(np.array([[ 1, 2, 3, 4],
... [ 5, 6, 7, 8],
... [10, 11, 12, 13],
... [14, 15, 16, 17]]))
>>> result = ops.tril(x)
>>> print(result)
[[ 1 0 0 0]
[ 5 6 0 0]
[10 11 12 0]
[14 15 16 17]]
>>> x = Tensor(np.array([[ 1, 2, 3, 4],
... [ 5, 6, 7, 8],
... [10, 11, 12, 13],
... [14, 15, 16, 17]]))
>>> result = ops.tril(x, diagonal=1)
>>> print(result)
[[ 1 2 0 0]
[ 5 6 7 0]
[10 11 12 13]
[14 15 16 17]]
>>> x = Tensor(np.array([[ 1, 2, 3, 4],
... [ 5, 6, 7, 8],
... [10, 11, 12, 13],
... [14, 15, 16, 17]]))
>>> result = ops.tril(x, diagonal=-1)
>>> print(result)
[[ 0 0 0 0]
[ 5 0 0 0]
[10 11 0 0]
[14 15 16 0]]
"""
tril_ = _get_cache_prim(Tril)(diagonal)
return tril_(input)
[docs]def tril_ext(input, diagonal=0):
"""
Returns the lower triangle part of `input` (elements that contain the diagonal and below),
and set the other elements to zeros.
Args:
input (Tensor): A Tensor with shape :math:`(x_1, x_2, ..., x_R)`. The rank must be at least 2.
Supporting all number types including bool.
diagonal (int, optional): An optional attribute indicates the diagonal to consider, default: 0,
indicating the main diagonal.
Returns:
Tensor, the same shape and data type as the `input`.
Raises:
TypeError: If `input` is not a Tensor.
TypeError: If `diagonal` is not an int.
TypeError: If the type of `input` is neither number nor bool.
ValueError: If the rank of `input` is less than 2.
Supported Platforms:
``Ascend``
Examples:
>>> import numpy as np
>>> from mindspore import Tensor, ops
>>> x = Tensor(np.array([[ 1, 2, 3, 4],
... [ 5, 6, 7, 8],
... [10, 11, 12, 13],
... [14, 15, 16, 17]]))
>>> result = ops.function.array_func.tril_ext(x)
>>> print(result)
[[ 1 0 0 0]
[ 5 6 0 0]
[10 11 12 0]
[14 15 16 17]]
>>> x = Tensor(np.array([[ 1, 2, 3, 4],
... [ 5, 6, 7, 8],
... [10, 11, 12, 13],
... [14, 15, 16, 17]]))
>>> result = ops.function.array_func.tril_ext(x, diagonal=1)
>>> print(result)
[[ 1 2 0 0]
[ 5 6 7 0]
[10 11 12 13]
[14 15 16 17]]
>>> x = Tensor(np.array([[ 1, 2, 3, 4],
... [ 5, 6, 7, 8],
... [10, 11, 12, 13],
... [14, 15, 16, 17]]))
>>> result = ops.function.array_func.tril_ext(x, diagonal=-1)
>>> print(result)
[[ 0 0 0 0]
[ 5 0 0 0]
[10 11 0 0]
[14 15 16 0]]
"""
return tril_ext_impl(input, diagonal)
@_primexpr
def _canonicalize_axis(axis, ndim):
"""
Check axes are within the number of dimensions of tensor x and normalize the negative axes.
Args:
axis (Union[int, tuple(int), list(int)]): Axes of the tensor.
ndim (int): The number of dimensions of the tensor.
Return:
Axis (Union[int, tuple(int)]). If input is integer, return integer, else tuple.
"""
if isinstance(axis, int):
axis = [axis]
for ax in axis:
if not isinstance(ax, int):
raise TypeError(f'axis should be integers, not {type(ax)}')
if not -ndim <= ax < ndim:
raise ValueError(
f'axis {ax} is out of bounds for array of dimension {ndim}')
def canonicalizer(ax):
return ax + ndim if ax < 0 else ax
axis = tuple([canonicalizer(ax) for ax in axis])
if all(axis.count(el) <= 1 for el in axis):
return tuple(sorted(axis)) if len(axis) > 1 else axis[0]
raise ValueError(f"duplicate axis in {axis}.")
@_primexpr
def _list_comprehensions(obj, item=None, return_tuple=False):
"""
Generates a new list or tuple by list comprehension.
Args:
obj (Union[int, list, tuple]):
If integer, it will be the length of the returned tuple/list.
item: The value to be filled. Default: ``None`` .
If ``None`` , the values in the new list/tuple are the same as obj
or range(obj) when obj is integer.
return_tuple(bool): If ``true`` , returns tuple, else returns list.
Returns:
List or tuple.
"""
lst = obj
if isinstance(obj, int):
lst = []
for i in ms_arrange(obj):
lst.append(i)
if item is None:
res = list(lst)
else:
res = [item for _ in lst]
if return_tuple:
return tuple(res)
return res
@_primexpr
def _tuple_setitem(tup, idx, value):
"""
Returns a tuple with specified `idx` set to `value`.
"""
tup = list(tup)
tup[idx] = value
return tuple(tup)
def _tensor_split_sub_tensors(x, indices_or_sections, axis):
"""
Splits the input tensor `x` into multiple sub-tensors along the axis according to the given `indices_or_sections`
with type of tuple or list.
"""
length_along_dim = x.shape[axis]
indices_or_sections = tuple(indices_or_sections)
indices_or_sections += (length_along_dim,)
sub_tensors = []
strides = _list_comprehensions(x.ndim, 1, True)
begin = _list_comprehensions(x.ndim, 0)
end = _list_comprehensions(x.shape)
for i in ms_arrange(len(indices_or_sections)):
idx = indices_or_sections[i]
begin[axis] = 0 if i == 0 else indices_or_sections[i - 1]
end[axis] = idx
sliced_tensor = strided_slice(x, tuple(begin), tuple(end), strides)
sub_tensors.append(sliced_tensor)
return tuple(sub_tensors)
def _tensor_split_sub_int(x, indices_or_sections, axis):
"""
Splits the input tensor `x` into multiple sub-tensors along the axis according to the given `indices_or_sections`
with type if int.
"""
arr_shape = x.shape
length_along_dim = arr_shape[axis]
if length_along_dim == 0:
res = _get_cache_prim(P.Split)(axis)(x)
elif indices_or_sections > length_along_dim:
res = _get_cache_prim(P.Split)(axis, length_along_dim)(x)
indices_or_sections_n = [length_along_dim, length_along_dim + 1]
res2 = _tensor_split_sub_tensors(x, indices_or_sections_n, axis)
for _ in np.arange(length_along_dim, indices_or_sections):
res += tuple(res2)[1:]
elif length_along_dim % indices_or_sections == 0:
res = _get_cache_prim(P.Split)(axis, indices_or_sections)(x)
else:
num_long_tensor = length_along_dim % indices_or_sections
num_short_tensor = indices_or_sections - num_long_tensor
length1 = num_long_tensor * \
(length_along_dim // indices_or_sections + 1)
length2 = length_along_dim - length1
start1 = _list_comprehensions(rank_(x), 0, True)
size1 = _tuple_setitem(arr_shape, axis, length1)
start2 = _tuple_setitem(start1, axis, length1)
size2 = _tuple_setitem(arr_shape, axis, length2)
res = _get_cache_prim(P.Split)(axis, num_long_tensor)(tensor_slice(x, start1, size1)) + \
_get_cache_prim(P.Split)(axis, num_short_tensor)(
tensor_slice(x, start2, size2))
return res
[docs]def tensor_split(input, indices_or_sections, axis=0):
r"""
Splits a tensor into multiple sub-tensors along the given axis.
Args:
input (Tensor): A Tensor to be divided.
indices_or_sections (Union[int, tuple(int), list(int)]):
- If `indices_or_sections` is an integer n, input tensor will be split into n sections.
- If :math:`input.shape[axis]` can be divisible by n, sub-sections will have equal size
:math:`input.shape[axis] / n` .
- If :math:`input.shape[axis]` is not divisible by n, the first :math:`input.shape[axis] \bmod n` sections
will have size :math:`input.shape[axis] // n + 1` , and the rest will have
size :math:`input.shape[axis] // n` .
- If `indices_or_sections` is of type tuple(int) or list(int), the input tensor will be split at the
indices in the list or tuple. For example, given parameters :math:`indices\_or\_sections=[1, 4]`
and :math:`axis=0` , the input tensor will be split into sections :math:`input[:1]` ,
:math:`input[1:4]` , and :math:`input[4:]` .
axis (int): The axis along which to split. Default: ``0`` .
Returns:
A tuple of sub-tensors.
Raises:
TypeError: If argument `input` is not Tensor.
TypeError: If argument `axis` is not int.
ValueError: If argument `axis` is out of range of :math:`[-input.ndim, input.ndim)` .
TypeError: If each element in 'indices_or_sections' is not integer.
TypeError: If argument `indices_or_sections` is not int, tuple(int) or list(int).
Supported Platforms:
``Ascend`` ``GPU`` ``CPU``
Examples:
>>> import numpy as np
>>> from mindspore import Tensor, ops
>>> input_x = np.arange(9).astype("float32")
>>> output = ops.tensor_split(Tensor(input_x), 3)
>>> print(output)
(Tensor(shape=[3], dtype=Float32, value= [ 0.00000000e+00, 1.00000000e+00, 2.00000000e+00]),
Tensor(shape=[3], dtype=Float32, value= [ 3.00000000e+00, 4.00000000e+00, 5.00000000e+00]),
Tensor(shape=[3], dtype=Float32, value= [ 6.00000000e+00, 7.00000000e+00, 8.00000000e+00]))
"""
if not isinstance(input, Tensor):
raise TypeError(f'expect `x` is a Tensor, but got {type(input)}')
if type(axis) is not int:
raise TypeError(
f"Type of Argument `axis` should be integer but got {type(axis)}")
handle_axis = _canonicalize_axis(axis, input.ndim)
if type(indices_or_sections) is int:
if indices_or_sections > 0:
res = _tensor_split_sub_int(
input, indices_or_sections, handle_axis)
else:
raise ValueError(f"For tensor_split, the value of 'indices_or_sections' must be more than zero "
f"but got {indices_or_sections}")
elif isinstance(indices_or_sections, (list, tuple)):
for item in indices_or_sections:
if type(item) is not int:
raise TypeError(
f"Each element in 'indices_or_sections' should be integer, but got {type(item)}.")
res = _tensor_split_sub_tensors(
input, indices_or_sections, handle_axis)
else:
raise TypeError(f"Type of Argument `indices_or_sections` should be integer, tuple(int) or list(int), "
f"but got {type(indices_or_sections)}")
return res
[docs]def vsplit(input, indices_or_sections):
"""
Splits `input` with two or more dimensions, into multiple sub-tensors vertically
according to `indices_or_sections`.
It is equivalent to `ops.tensor_split` with :math:`axis=0` .
Args:
input (Tensor): A Tensor to be divided.
indices_or_sections (Union[int, tuple(int), list(int)]): See argument in :func:`mindspore.ops.tensor_split`.
Returns:
A list of sub-tensors.
Supported Platforms:
``Ascend`` ``GPU`` ``CPU``
Examples:
>>> import numpy as np
>>> from mindspore import Tensor, ops
>>> input_x = np.arange(9).reshape((3, 3)).astype('float32')
>>> output = ops.vsplit(Tensor(input_x), 3)
>>> print(output)
(Tensor(shape=[1, 3], dtype=Float32, value=[[ 0.00000000e+00, 1.00000000e+00, 2.00000000e+00]]),
Tensor(shape=[1, 3], dtype=Float32, value=[[ 3.00000000e+00, 4.00000000e+00, 5.00000000e+00]]),
Tensor(shape=[1, 3], dtype=Float32, value=[[ 6.00000000e+00, 7.00000000e+00, 8.00000000e+00]]))
"""
if not isinstance(input, Tensor):
raise TypeError(f'expect `x` is a Tensor, but got {type(input)}')
if input.ndim < 1:
raise ValueError(
f'vsplit expect `x` is a Tensor with at least 1 dimension, but got {input.ndim}')
return tensor_split(input, indices_or_sections, 0)
[docs]def hsplit(input, indices_or_sections):
"""
Splits a tensor into multiple sub-tensors horizontally.
It is equivalent to `ops.tensor_split` with :math:`axis=1` .
Args:
input (Tensor): A Tensor to be divided.
indices_or_sections (Union[int, tuple(int), list(int)]): See argument in :func:`mindspore.ops.tensor_split`.
Returns:
A list of sub-tensors.
Raises:
TypeError: If `input` is not Tensor.
ValueError: If dimension of `input` is less than 2.
Supported Platforms:
``Ascend`` ``GPU`` ``CPU``
Examples:
>>> import numpy as np
>>> from mindspore import Tensor, ops
>>> input_x = np.arange(6).reshape((2, 3)).astype('float32')
>>> output = ops.hsplit(Tensor(input_x), 3)
>>> print(output)
(Tensor(shape=[2, 1], dtype=Float32, value=[[ 0.00000000e+00], [ 3.00000000e+00]]),
Tensor(shape=[2, 1], dtype=Float32, value=[[ 1.00000000e+00], [ 4.00000000e+00]]),
Tensor(shape=[2, 1], dtype=Float32, value=[[ 2.00000000e+00], [ 5.00000000e+00]]))
"""
if not isinstance(input, Tensor):
raise TypeError(f'expect `x` is a Tensor, but got {type(input)}')
if input.ndim < 2:
raise ValueError(
f'hsplit expect `x` is a Tensor with at least 2 dimension, but got {input.ndim}')
return tensor_split(input, indices_or_sections, 1)
[docs]def dsplit(input, indices_or_sections):
"""
Splits a tensor into multiple sub-tensors along the 3rd axis.
It is equivalent to `ops.tensor_split` with :math:`axis=2` .
Args:
input (Tensor): A Tensor to be divided.
indices_or_sections (Union[int, tuple(int), list(int)]): See argument in :func:`mindspore.ops.tensor_split`.
Returns:
A list of sub-tensors.
Supported Platforms:
``Ascend`` ``GPU`` ``CPU``
Examples:
>>> import numpy as np
>>> from mindspore import Tensor, ops
>>> input_x = np.arange(6).reshape((1, 2, 3)).astype('float32')
>>> output = ops.dsplit(Tensor(input_x), 3)
>>> print(output)
(Tensor(shape=[1, 2, 1], dtype=Float32, value=[[[ 0.00000000e+00], [ 3.00000000e+00]]]),
Tensor(shape=[1, 2, 1], dtype=Float32, value=[[[ 1.00000000e+00], [ 4.00000000e+00]]]),
Tensor(shape=[1, 2, 1], dtype=Float32, value=[[[ 2.00000000e+00], [ 5.00000000e+00]]]))
"""
if not isinstance(input, Tensor):
raise TypeError(f'expect `x` is a Tensor, but got {type(input)}')
if input.ndim < 3:
raise ValueError(
f'dsplit expect `x` is a Tensor with at least 3 dimension, but got {input.ndim}')
return tensor_split(input, indices_or_sections, 2)
def _init_and_select_elem(input, initial, where, cmp_fn): # pylint: disable=redefined-outer-name
"""Initialize the input according to Initial, and select the element according to where."""
if initial is not None:
initial = ops.fill(input.dtype, input.shape, initial)
input = cmp_fn(input, initial)
if where is not None and not isinstance(where, Tensor):
where = Tensor(where, dtype=mstype.bool_)
if where is not None and (where.shape or not where):
if initial is None:
raise ValueError('initial value must be provided for where masks')
where = where.broadcast_to(input.shape)
initial = initial.broadcast_to(input.shape)
input = ops.select(where, input, initial)
return input
[docs]def max(input, axis=None, keepdims=False, *, initial=None, where=None): # pylint: disable=redefined-outer-name
"""
Calculates the maximum value along with the given axis for the input tensor. It returns the maximum values and
indices.
Note:
- In auto_parallel and semi_auto_parallel mode, the first output index can not be used.
- When `axis` is ``None``, `keepdims` and subsequent parameters have no
effect. At the same time, the index is fixed to return 0.
.. warning::
- If there are multiple maximum values, the index of the first maximum value is used.
- The value range of "axis" is [-dims, dims - 1]. "dims" is the dimension length of "input".
Also see: :class:`mindspore.ops.ArgMaxWithValue`.
Args:
input (Tensor): The input tensor, can be any dimension. Complex tensor is not supported for now.
axis (int): The dimension to reduce. When `axis` is ``None``, computing the maximum value of all elements
in `input` .Default: ``None`` .
keepdims (bool): Whether to reduce dimension, if true, the output will keep same dimension with the input,
the output will reduce dimension if false. Default: ``False`` .
Keyword Args:
initial (scalar, optional): The minimum value of an output element. Must be present to allow computation
on empty slice. Default: ``None`` .
where (Tensor[bool], optional): A Tensor indicating whether to replace the primitive value in `input`
with the value in `initial`. If ``True`` , do not replace, otherwise replace. For the index of ``True``
in `where`, the corresponding value in `initial` must be assigned. Default: ``None`` , which indicates
``True`` by default.
Returns:
tuple (Tensor), tuple of 2 tensors, containing the corresponding index and the maximum value of the input
tensor.
- values (Tensor) - The maximum value of input tensor, with the same shape as index, and same dtype as x.
- index (Tensor) - The index for the maximum value of the input tensor, with dtype int64. If `keepdims`
is true, the shape of output tensors is :math:`(input_1, input_2, ..., input_{axis-1}, 1, input_{axis+1},
..., input_N)` . Otherwise, the shape is :math:`(input_1, input_2, ..., input_{axis-1}, input_{axis+1},
..., input_N)` .
Raises:
TypeError: If `input` is not Tensor.
TypeError: If `keepdims` is not a bool.
TypeError: If `axis` is not an int.
TypeError: If `initial` is not a number.
Supported Platforms:
``Ascend`` ``GPU`` ``CPU``
Examples:
>>> import mindspore
>>> import numpy as np
>>> from mindspore import Tensor, ops
>>> x = Tensor(np.array([0.0, 0.4, 0.6, 0.7, 0.1]), mindspore.float32)
>>> output, index = ops.max(x)
>>> print(output, index)
0.7 0
>>> y = Tensor(np.array([[0.0, 0.3, 0.4, 0.5, 0.1],
... [3.2, 0.4, 0.1, 2.9, 4.0]]), mindspore.float32)
>>> output, index = ops.max(y, axis=0, keepdims=True)
>>> print(output, index)
[[3.2 0.4 0.4 2.9 4. ]] [[1 1 0 1 1]]
"""
if not input.shape:
return (input, Tensor(0, dtype=mstype.int64))
if axis is None:
return (max_(input), Tensor(0, dtype=mstype.int64))
if initial is not None and not isinstance(initial, numbers.Number):
raise TypeError(
f"For 'max', 'initial' must be a scalar, but got {type(initial)}")
if axis is not None and not isinstance(axis, int):
raise TypeError(f"For 'max', 'axis' must be int, but got {type(axis)}")
input = _init_and_select_elem(input, initial, where, ops.maximum)
argmax_with_value_op = _get_cache_prim(ArgMaxWithValue)(axis, keepdims)
indices, values = argmax_with_value_op(input)
return values, indices
[docs]def argmax(input, dim=None, keepdim=False):
"""
Return the indices of the maximum values of a tensor across a dimension.
Args:
input (Tensor): Input tensor.
dim (Union[int, None], optional): The dimension to reduce. If `dim` is ``None`` , the indices of the maximum
value within the flattened input will be returned. Default: ``None`` .
keepdim (bool, optional): Whether the output tensor retains the specified
dimension. Ignored if `dim` is None. Default: ``False`` .
Returns:
Tensor, indices of the maximum values across a dimension.
Raises:
TypeError: If `keepdim` is not bool.
ValueError: If `dim` is out of range.
Supported Platforms:
``Ascend`` ``GPU`` ``CPU``
Examples:
>>> import numpy as np
>>> from mindspore import Tensor, ops
>>> x = Tensor(np.array([[1, 20, 5], [67, 8, 9], [130, 24, 15]]).astype(np.float32))
>>> output = ops.argmax(x, dim=-1)
>>> print(output)
[1 0 0]
"""
_check_attr_dtype("keepdim", keepdim, [bool], "argmax")
if not input.shape:
return Tensor(0)
if input.dtype == mstype.bool_:
input = input.astype(mstype.int32)
is_dim_none = False
if dim is None:
input = reshape_(input, (-1,))
dim = 0
is_dim_none = True
out = _get_cache_prim(Argmax)(dim, mstype.int64)(input)
if keepdim and not is_dim_none:
out = expand_dims(out, dim)
return out
[docs]def min(input, axis=None, keepdims=False, *, initial=None, where=None): # pylint: disable=redefined-outer-name
"""
Calculates the minimum value along with the given axis for the input tensor. It returns the minimum values and
indices.
Note:
- In auto_parallel and semi_auto_parallel mode, the first output index can not be used.
- When `axis` is ``None``, `keepdims` and subsequent parameters have no
effect. At the same time, the index is fixed to return 0.
.. warning::
- If there are multiple minimum values, the index of the first minimum value is used.
- The value range of "axis" is [-dims, dims - 1]. "dims" is the dimension length of "x".
Args:
input (Tensor): The input tensor, can be any dimension. Complex tensor is not supported for now.
axis (int): The dimension to reduce. Default: ``None`` .
keepdims (bool): Whether to reduce dimension, if ``True`` the output will keep the same dimension as the input,
the output will reduce dimension if ``False`` . Default: ``False`` .
Keyword Args:
initial (scalar, optional): The maximum value of an output element. Must be present to allow computation
on empty slice. Default: ``None`` .
where (Tensor[bool], optional): A Tensor indicating whether to replace the primitive value in `input`
with the value in `initial`. If ``True`` , do not replace, otherwise replace. For the index of ``True``
in `where`, the corresponding value in `initial` must be assigned. Default: ``None`` , which indicates
``True`` by default.
Returns:
tuple (Tensor), tuple of 2 tensors, containing the corresponding index and the minimum value of the input
tensor.
- **values** (Tensor) - The minimum value of input tensor, with the same
shape as `index`, and same dtype as `x`.
- **index** (Tensor) - The index for the minimum value of the input tensor, with dtype int32. If `keepdims`
is true, the shape of output tensors is :math:`(input_1, input_2, ..., input_{axis-1}, 1, input_{axis+1},
..., input_N)` . Otherwise, the shape is :math:`(input_1, input_2, ..., input_{axis-1}, input_{axis+1},
..., input_N)` .
Raises:
TypeError: If `x` is not Tensor.
TypeError: If `keepdims` is not a bool.
TypeError: If `axis` is not an int.
TypeError: If `initial` is not a number.
Supported Platforms:
``Ascend`` ``GPU`` ``CPU``
Examples:
>>> import mindspore
>>> import numpy as np
>>> from mindspore import Tensor, ops
>>> x = Tensor(np.array([0.0, 0.4, 0.6, 0.7, 0.1]), mindspore.float32)
>>> output, index = ops.min(x, keepdims=True)
>>> print(output, index)
0.0 0
"""
if not input.shape:
return (input, Tensor(0, dtype=mstype.int64))
if axis is None:
return (min_(input), Tensor(0, dtype=mstype.int64))
if initial is not None and not isinstance(initial, numbers.Number):
raise TypeError(
f"For 'min', 'initial' must be a scalar, but got {type(initial)}")
if axis is not None and not isinstance(axis, int):
raise TypeError(f"For 'min', 'axis' must be int, but got {type(axis)}")
input = _init_and_select_elem(input, initial, where, ops.minimum)
argmin_with_value_op = _get_cache_prim(ArgMinWithValue)(axis, keepdims)
indices, values = argmin_with_value_op(input)
return values, indices
[docs]def aminmax(input, *, axis=0, keepdims=False):
"""
It returns the minimum and maximum value along the given axis of input tensor.
Args:
input (Tensor): The input tensor, can be any dimension. Set the shape of input tensor as
:math:`(x_1, x_2, ..., x_N)` .
Keyword Args:
axis (int, optional): The dimension to reduce. The value range of `axis` is [-rank, rank),
where "rank" is the dimension of `input`. If `axis` is None, computes the minimum and maximum value
along the entire input tensor. Default: ``0`` .
keepdims (bool, optional): Whether to maintain dimension. When set to True, the output will keep the same
dimension as the input, or the dimension specified by `axis` is reduced. Default: ``False`` .
Returns:
tuple (Tensor), containing the minimum value and maximum value of the input tensor.
- If `keepdims` is True, the shape of output tensors is
:math:`(x_1, x_2, ..., x_{axis-1}, 1, x_{axis+1}, ..., x_N)`.
- If `keepdims` is False, the shape of output tensors is
:math:`(x_1, x_2, ..., x_{axis-1}, x_{axis+1}, ..., x_N)`.
Raises:
TypeError: If `keepdims` is not a bool.
TypeError: If `axis` is not an int and not None.
ValueError: If `axis` is not in range [-rank, rank).
Supported Platforms:
``Ascend`` ``GPU`` ``CPU``
Examples:
>>> import mindspore
>>> import numpy as np
>>> from mindspore import Tensor, ops
>>> x = Tensor(np.array([0.0, 0.4, 0.6, 0.7, 0.1]), mindspore.float32)
>>> output0, output1 = ops.aminmax(x)
>>> print(output0, output1)
0.0 0.7
>>> output2, output3 = ops.aminmax(x, axis=-1, keepdims=True)
>>> print(output2, output3)
[0.] [0.7]
>>> x = Tensor(np.array([[0.0, 0.4, 0.6, 0.7, 0.1], [0.78, 0.97, 0.5, 0.82, 0.99]]), mindspore.float32)
>>> output4, output5 = ops.aminmax(x, axis=None, keepdims=True)
>>> print(output4, output5)
[[0.]] [[0.99]]
"""
if axis is None:
output0, _ = ops.min(input, axis, keepdims)
output1, _ = ops.max(input, axis, keepdims)
if keepdims is True:
output0 = ops.reshape(output0, [1] * input.ndim)
output1 = ops.reshape(output1, [1] * input.ndim)
return output0, output1
argmin_with_value_op = _get_cache_prim(ArgMinWithValue)(axis, keepdims)
argmax_with_value_op = _get_cache_prim(ArgMaxWithValue)(axis, keepdims)
_, output0 = argmin_with_value_op(input)
_, output1 = argmax_with_value_op(input)
if keepdims is True and input.ndim == 0:
output0 = ops.reshape(output0, [1])
output1 = ops.reshape(output1, [1])
return output0, output1
[docs]def narrow(input, axis, start, length):
"""
Returns a narrowed tensor from input tensor, and
the dimension axis is input from start to start + length.
Args:
input (Tensor): the tensor to narrow.
axis (int): the axis along which to narrow.
start (int): the starting dimension.
length (int): the distance to the ending dimension.
Returns:
Tensor.
- output (Tensors) - The narrowed tensor.
Raises:
TypeError: If the input is not a tensor or tuple or list of tensors.
Supported Platforms:
``Ascend`` ``GPU`` ``CPU``
Examples:
>>> import mindspore
>>> from mindspore import ops
>>> from mindspore import Tensor
>>> x = Tensor([[1, 2, 3], [4, 5, 6], [7, 8, 9]], mindspore.int32)
>>> output = ops.narrow(x, 0, 0, 2)
>>> print(output)
[[ 1 2 3]
[ 4 5 6]]
>>> output = ops.narrow(x, 1, 1, 2)
>>> print(output)
[[ 2 3]
[ 5 6]
[ 8 9]]
"""
validator.check_value_type("input", input, Tensor, "narrow")
validator.check_axis_in_range(axis, input.ndim)
validator.check_int_range(start, 0, input.shape[axis], validator.INC_LEFT)
validator.check_int_range(
length, 1, input.shape[axis] - start, validator.INC_BOTH)
begins = [0] * input.ndim
begins[axis] = start
sizes = list(input.shape)
sizes[axis] = length
return tensor_slice(input, begins, sizes)
[docs]def narrow_ext(input, dim, start, length):
"""
Returns a narrowed tensor from input tensor, and
the dimension axis is input from start to start + length.
Args:
input (Tensor): the tensor to narrow.
dim (int): dimension along which to narrow.
start (int): the starting dimension.
length (int): the distance to the ending dimension.
Returns:
Tensor.
Raises:
ValueError: If dim is out of range [-input.ndim, input.ndim).
ValueError: If start is out of range [-input.shape[dim], input.shape[dim]].
ValueError: It length is out of range [0, input.shape[dim]-start].
Supported Platforms:
``Ascend``
Examples:
>>> import mindspore
>>> from mindspore import ops
>>> from mindspore import Tensor
>>> x = Tensor([[1, 2, 3], [4, 5, 6], [7, 8, 9]], mindspore.int32)
>>> output = ops.narrow(x, 0, 0, 2)
>>> print(output)
[[ 1 2 3]
[ 4 5 6]]
>>> output = ops.narrow(x, 1, 1, 2)
>>> print(output)
[[ 2 3]
[ 5 6]
[ 8 9]]
"""
validator.check_value_type("input", input, Tensor, "narrow")
return slice_ext_op(input, dim, start, start + length, 1)
[docs]def topk(input, k, dim=None, largest=True, sorted=True):
r"""
Finds values and indices of the `k` largest or smallest entries along a given dimension.
.. warning::
- If sorted is set to False, it will use the aicpu operator, the performance may be reduced. In addition, due to
different memory layout and traversal methods on different platforms, the display order of calculation results
may be inconsistent when `sorted` is False.
If the `input` is a one-dimensional Tensor, finds the `k` largest or smallest entries in the Tensor,
and outputs its value and index as a Tensor. values[`k`] is the `k` largest item in `input`,
and its index is indices [`k`].
For a multi-dimensional matrix,
calculates the first or last `k` entries in a given dimension, therefore:
.. math::
values.shape = indices.shape
If the two compared elements are the same, the one with the smaller index value is returned first.
Args:
input (Tensor): Input to be computed, data type must be float16, float32 or int32.
k (int): The number of top or bottom elements to be computed along the last dimension.
dim (int, optional): The dimension to sort along. Default: ``None`` .
largest (bool, optional): If largest is ``False`` then the k smallest elements are returned.
Default: ``True`` .
sorted (bool, optional): If ``True`` , the obtained elements will be sorted by the values in descending order.
If ``False`` , the obtained elements will not be sorted. Default: ``True`` .
Returns:
A tuple consisting of `values` and `indexes`.
- values (Tensor): The `k` largest or smallest elements in each slice of the given dimension.
- indices (Tensor): The indices of values within the last dimension of input.
Raises:
TypeError: If `sorted` is not a bool.
TypeError: If `input` is not a Tensor.
TypeError: If `k` is not an int.
TypeError: If dtype of `input` is not one of the following: float16, float32 or int32.
Supported Platforms:
``Ascend`` ``GPU`` ``CPU``
Examples:
>>> import mindspore as ms
>>> from mindspore import ops
>>> x = ms.Tensor([[0.5368, 0.2447, 0.4302, 0.9673],
... [0.4388, 0.6525, 0.4685, 0.1868],
... [0.3563, 0.5152, 0.9675, 0.8230]], dtype=ms.float32)
>>> output = ops.topk(x, 2, dim=1)
>>> print(output)
(Tensor(shape=[3, 2], dtype=Float32, value=
[[ 9.67299998e-01, 5.36800027e-01],
[ 6.52499974e-01, 4.68499988e-01],
[ 9.67499971e-01, 8.23000014e-01]]), Tensor(shape=[3, 2], dtype=Int32, value=
[[3, 0],
[1, 2],
[2, 3]]))
>>> output2 = ops.topk(x, 2, dim=1, largest=False)
>>> print(output2)
(Tensor(shape=[3, 2], dtype=Float32, value=
[[ 2.44700000e-01, 4.30200011e-01],
[ 1.86800003e-01, 4.38800007e-01],
[ 3.56299996e-01, 5.15200019e-01]]), Tensor(shape=[3, 2], dtype=Int32, value=
[[1, 2],
[3, 0],
[0, 1]]))
"""
top_k_ = _get_cache_prim(P.TopK)(sorted)
if not largest:
input = -input
if dim is None or dim == input.ndim - 1:
if not largest:
res = top_k_(input, k)
values, indices = -res[0], res[1]
return values, indices
return top_k_(input, k)
input = input.swapaxes(dim, input.ndim - 1)
output = top_k_(input, k)
values = output[0].swapaxes(dim, input.ndim - 1)
indices = output[1].swapaxes(dim, input.ndim - 1)
if not largest:
res = (-values, indices)
else:
res = (values, indices)
return res
def expand(input_x, size):
r"""
:func:`mindspore.ops.expand` will be deprecated in the future.
Please use :func:`mindspore.ops.broadcast_to` instead.
"""
expand_op = _get_cache_prim(Expand)()
return expand_op(input_x, size)
@_primexpr
def _check_fold_param(param, param_name):
"""Check the parameters of fold op."""
validator.check_value_type(param_name, param, [int, list, tuple], 'fold')
param = (param, param) if isinstance(param, int) else param
validator.check_int(len(param), 2, validator.EQ, param_name, 'fold')
if param_name == "padding":
validator.check_non_negative_int_sequence(param, param_name, 'fold')
else:
validator.check_positive_int_sequence(param, param_name, 'fold')
return param
[docs]def fold(input, output_size, kernel_size, dilation=1, padding=0, stride=1):
r"""
Combines an array of sliding local blocks into a large containing tensor.
Consider a batched input tensor of shape :math:`(N, C \times \prod(\text{kernel_size}), L)` ,
where :math:`N` is the batch dimension, :math:`C \times \prod(\text{kernel_size})` is the
total number of values within each block (a block has :math:`\prod(\text{kernel_size})` spatial
locations each containing a `C`-channeled vector), and :math:`L` is the total number of such blocks:
.. math::
L = \prod_d \left\lfloor\frac{\text{output_size}[d] + 2 \times \text{padding}[d] %
- \text{dilations}[d] \times (\text{kernel_size}[d] - 1) - 1}{\text{strides}[d]} + 1\right\rfloor,
where :math:`d` is over all spatial dimensions.
Therefore, `output_size` is the spatial shape of the large containing tensor of the sliding local blocks.
The `dilation`, `padding` and `stride` arguments specify how the sliding blocks are retrieved.
.. warning::
- The input must be a 3-dimensional Tensor with shape
:math:`(N, C \times \prod(\text{kernel_size}), L)` .
- The output must be a 4-dimensional Tensor with shape
:math:`(N, C, output\_size[0], output\_size[1], ...)` .
Args:
input (Tensor): 3-D Tensor, supported dtypes: float16, float32, float64, complex64 and complex128.
output_size (Tensor): 1D tensor with `2` elements of data type int.
kernel_size (Union[int, tuple[int], list[int]]): The size of the kernel, should be two int
for height and width. If type is int, it means that height equal with width. Must be specified.
dilation (Union[int, tuple[int], list[int]], optional): The size of the dilation, should be two int
for height and width. If type is int, it means that height equal with width. Default: ``1`` .
padding (Union[int, tuple[int], list[int]], optional): The size of the padding, should be two int
for height and width. If type is int, it means that height equal with width. Default: ``0`` .
stride (Union[int, tuple[int], list[int]], optional): The size of the stride, should be two int
for height and width. If type is int, it means that height equal with width. Default: ``1`` .
Returns:
A Tensor, with same type as `input` . And its shape is as described above.
Raises:
TypeError: If `output_size`, `kernel_size`, `stride`, `dilation`, `padding` data type is not int, tuple or list.
ValueError: If `output_size`, `kernel_size`, `dilation`, `stride` value is not
greater than zero or elements number more than `2`.
ValueError: If `padding` value is less than zero or elements number more than `2`.
ValueError: If `input.shape[1] != kernel_size[0] * kernel_size[1]`
ValueError: If `input.shape[2]` does not match the calculated number of sliding blocks.
Supported Platforms:
``Ascend`` ``GPU`` ``CPU``
Examples:
>>> import numpy as np
>>> from mindspore import Tensor, ops
>>> from mindspore import dtype as mstype
>>> x = Tensor(input_data=np.random.rand(16, 64, 25), dtype=mstype.float32)
>>> output_size = Tensor(input_data=[8, 8], dtype=mstype.int32)
>>> output = ops.fold(x, output_size, [2, 2], [2, 2], [2, 2], [2, 2])
>>> print(output.shape)
(16, 16, 8, 8)
"""
kernel_size = _check_fold_param(kernel_size, "kernel_size")
dilation = _check_fold_param(dilation, "dilation")
padding = _check_fold_param(padding, "padding")
stride = _check_fold_param(stride, "stride")
fold_op = _get_cache_prim(Col2Im)(kernel_size, dilation, padding, stride)
input_shape = ops.shape(input)
k = kernel_size[0] * kernel_size[-1]
r_shape = input_shape[:1] + (-1, k) + input_shape[-1:]
input = ops.reshape(input, r_shape)
return fold_op(input, output_size)
@_primexpr
def _check_unfold_params(param, param_name, param_size):
"""Check the parameters of unfold op."""
validator.check_value_type(param_name, param, [int, tuple, list], 'unfold')
param = (param, param) if isinstance(param, int) else param
validator.check(param_name + " size", len(param), "",
param_size, validator.IN, 'unfold')
if param_name == "padding":
validator.check_non_negative_int_sequence(param, param_name, 'unfold')
else:
validator.check_positive_int_sequence(param, param_name, 'unfold')
return param
[docs]def unfold(input, kernel_size, dilation=1, padding=0, stride=1):
r"""
Extracts sliding local blocks from a batched input tensor.
Consider a batched input tensor of shape :math:`(N, C, *)`,
where :math:`N` is the batch dimension, :math:`C` is the channel dimension,
and :math:`*` represent arbitrary spatial dimensions. This operation flattens
each sliding `Kernel_size`- sized block within the spatial dimensions
of input `x` into a column (i.e., last dimension) of a 3-D output
tensor of shape :math:`(N, C \times \prod(\text{kernel_size}), L)`, where
:math:`C \times \prod(\text{kernel_size})` is the total number of values
within each block (a block has :math:`\prod(\text{kernel_size})` spatial
locations each containing a `C`-channeled vector), and :math:`L` is
the total number of such blocks:
.. math::
L = \prod_d \left\lfloor\frac{\text{spatial_size}[d] + 2 \times \text{pads}[d] %
- \text{dilations}[d] \times (\text{kernel_size}[d] - 1) - 1}{\text{strides}[d]} + 1\right\rfloor,
where :math:`\text{spatial_size}` is formed by the spatial dimensions
of input `x` (:math:`*` above), and :math:`d` is over all spatial
dimensions.
Therefore, indexing `output` at the last dimension (column dimension)
gives all values within a certain block.
The `dilation`, `padding` and `stride` arguments specify
how the sliding blocks are retrieved.
.. warning::
- The output is a 3-dimensional Tensor whose shape is
:math:`(N, C \times \prod(\text{kernel_size}), L)` .
- This is an experimental API that is subject to change or deletion.
Args:
input (Tensor): 4-D Tensor, supported dtypes: float16, float32, float64, complex64 and complex128.
kernel_size (Union[int, tuple[int], list[int]]): The size of the kernel, should be two int
for height and width. If type is int, it means that height equal with width. Must be specified.
dilation (Union[int, tuple[int], list[int]], optional): The dilation of the window, should be two int
for height and width. If type is int, it means that height equal with width. Default: ``1`` .
padding (Union[int, tuple[int], list[int]], optional): The pad of the window, that must be
a tuple/list of one or two `int` for height and width. Default: ``0`` .
- If one int, pad_height = pad_width.
- If two int, pad_height = padding[0], pad_width = padding[1].
stride (Union[int, tuple[int], list[int]], optional): The stride of the window, should be two int
for height and width. If type is int, it means that height equal with width. Default: ``1`` .
Returns:
A Tensor, with same type as `input` . And its shape is as described above.
Raises:
TypeError: If any data type of `kernel_size`, `stride`, `dilation`, `padding` is not int, tuple or list.
ValueError: If `kernel_size`, `dilation`, `stride` value is not
greater than zero or elements number more than `2`.
ValueError: If `padding` value is less than zero.
Supported Platforms:
``Ascend`` ``GPU`` ``CPU``
Examples:
>>> import mindspore
>>> import numpy as np
>>> from mindspore import Tensor, ops
>>> x = Tensor(np.random.rand(4, 4, 32, 32), mindspore.float64)
>>> output = ops.unfold(x, kernel_size=3, dilation=1, stride=1)
>>> print(output.shape)
(4, 36, 900)
"""
kernel_size = _check_unfold_params(kernel_size, "kernel_size", [1, 2])
dilation = _check_unfold_params(dilation, "dilation", [1, 2])
padding = _check_unfold_params(padding, "padding", [1, 2])
stride = _check_unfold_params(stride, "stride", [1, 2])
unfold_op = _get_cache_prim(Im2Col)(ksizes=kernel_size,
strides=stride,
dilations=dilation,
pads=padding)
tmp = unfold_op(input)
tmp_shape = ops.shape(tmp)
out_shape = tmp_shape[:1] + (-1,) + tmp_shape[-1:]
out = ops.reshape(tmp, out_shape)
return out
@_primexpr
def _check_diagonal_axes(dim1, dim2, x_ndim):
"""Check the parameters of unfold op."""
axes = validator.check_axis_valid((dim1, dim2), x_ndim)
return axes
def _check_is_tensor(param_name, input, cls_name):
"""Returns True if input is Tensor."""
if not isinstance(input, Tensor):
raise TypeError(
f"For {cls_name}, {param_name} must be a Tensor, but got {type(input)}.")
@_primexpr
def _check_diagonal_scatter_shape(diag_shape, src_shape):
if diag_shape != src_shape:
raise ValueError(f"For diagonal_scatter, the shape of src should equal to the shape of input diagonal,"
f"but got src.shape {src_shape} and diagonal shape {diag_shape}.")
[docs]def diagonal_scatter(input, src, offset=0, dim1=0, dim2=1):
"""
`dim1` and `dim2` specify the two dimensions of `input`,
the elements in these two dimensions will be treated as elements of a matrix,
and `src` is embedded on the diagonal of the matrix.
Note:
Currently, ``inf`` value of elements in `input` or `src` is not supported.
Args:
input (Tensor): Input Tensor, whose dimension is larger than 1.
src (Tensor): The source Tensor to embed.
offset (int, optional): `offset` controls which diagonal to choose. Default: ``0`` .
- When `offset` is zero, the diagonal chosen is the main diagonal.
- When `offset` is a positive integer, the diagonal chosen is up the main diagonal.
- When `offset` is a negative integer, the diagonal chosen is down the main diagonal.
dim1 (int, optional): Axis to be used as the first axis of the 2-D
sub-arrays from which the diagonals should be taken. Default: ``0`` .
dim2 (int, optional): Axis to be used as the second axis of the 2-D
sub-arrays from which the diagonals should be taken. Default: ``1`` .
Returns:
Tensor after embedding, has the same shape and dtype as `input`.
Raises:
TypeError: If `input` or `src` is not a Tensor.
TypeError: If `offset` , `dim1` or `dim2` is not an integer.
Supported Platforms:
``Ascend`` ``GPU`` ``CPU``
Examples:
>>> import mindspore as ms
>>> input = ms.ops.zeros((3,3))
>>> src = ms.ops.ones(2)
>>> out = ms.ops.diagonal_scatter(input, src, 1, dim1=1, dim2=0)
>>> print(out)
[[0. 0. 0.]
[1. 0. 0.]
[0. 1. 0.]]
"""
_check_is_tensor("input", input, "diagonal_scatter")
_check_is_tensor("src", src, "diagonal_scatter")
input_diag = input.diagonal(offset, dim1, dim2)
_check_diagonal_scatter_shape(input_diag.shape, src.shape)
input_shape = input.shape
zeros_shape = list(input_shape)
m, n = input_shape[dim1], input_shape[dim2]
if m == n:
src = src - input_diag
src = ops.diag_embed(src, offset, dim1, dim2)
return input + src
if m > n:
axis = dim2
zeros_shape[axis] = m - n
else:
axis = dim1
zeros_shape[axis] = n - m
zeros_tensor = zeros(zeros_shape, dtype=input.dtype)
input = concat((input, zeros_tensor), axis)
input_diag = input.diagonal(offset, dim1, dim2)
if src.shape != input_diag.shape:
zeros_shape = []
for i, ax in enumerate(src.shape):
if ax == input_diag.shape[i]:
zeros_shape.append(ax)
else:
axis = i
zeros_shape.append(input_diag.shape[i] - ax)
zeros_tensor = zeros(zeros_shape, dtype=src.dtype)
src = concat((src, zeros_tensor), axis)
src = src - input_diag
src = ops.diag_embed(src, offset, dim1, dim2)
input = input + src
begin = (0,) * input.ndim
return slice(input, begin, input_shape)
def lstsq(input, A):
r"""
Computes the solutions of the least squares and minimum norm problems of full-rank
matrix `x` of size :math:`(m \times n)` and matrix `a` of size :math:`(m \times k)`.
If :math:`m \geq n`, `lstsq` solves the least-squares problem:
.. math::
\begin{array}{ll}
\min_y & \|xy-a\|_2.
\end{array}
If :math:`m < n`, `lstsq` solves the least-norm problem:
.. math::
\begin{array}{llll}
\min_y & \|y\|_2 & \text{subject to} & xy = a.
\end{array}
where `y` is the returned tensor.
Args:
input (Tensor): The :math:`(m \times n)` matrix equivalent to :math:`x` in above.
The input tensor whose data type is float16, float32 or float64.
A (Tensor): The :math:`(m \times k)` matrix equivalent to :math:`a` in above.
The input tensor whose data type is float16, float32 or float64.
Returns:
Tensor, the least squares or minimum norm problems solution, which has shape :math:`(n \times k)`.
The data type is the same with `input`.
Raises:
TypeError: If `input` or `A` is not a Tensor.
TypeError: If dtype of `input` or `A` is not one of: float16, float32, float64.
TypeError: If the dtypes of `input` and `A` are not the same.
ValueError: If the dimension of `input` is not equal to 2.
ValueError: If the dimension of `A` is not equal to 2 or 1.
ValueError: If the length of input_dims[0] is not equal to the length of A_dims[0].
Supported Platforms:
``CPU``
Examples:
>>> import mindspore
>>> import numpy as np
>>> from mindspore import Tensor, ops
>>> x = Tensor(np.array([[2,1,5],[3,5,1],[1,1,1]]),mindspore.float32)
>>> a = Tensor(np.array([[10,5],[15,8],[7,4]]),mindspore.float32)
>>> output = ops.lstsq(x, a)
>>> print(output)
[[17.000002 11.000002 ]
[-6.5000005 -4.500001 ]
[-3.500002 -2.5000017]]
"""
return lstsq_(input, A)
[docs]def mvlgamma(input, p):
r"""
Returns the results of the multivariate log-gamma function with dimension `p` element-wise.
The mathematical calculation process of Mvlgamma is shown as follows:
.. math::
\log (\Gamma_{p}(input))=C+\sum_{i=1}^{p} \log (\Gamma(input-\frac{i-1}{2}))
where :math:`C = \log(\pi) \times \frac{p(p-1)}{4}` and :math:`\Gamma(\cdot)` is the Gamma function.
Args:
input (Tensor): The input tensor of the multivariate log-gamma function,
which must be one of the following types: float32, float64.
The shape is :math:`(N,*)`, where :math:`*` means any number of additional dimensions.
And the value of any element in `input` must be greater than :math:`(p - 1) / 2`.
p (int): The number of dimensions. And the value of `p` must be greater than or equal to 1.
Returns:
Tensor, has the same shape and type as `input`.
Raises:
TypeError: If dtype of `input` is neither float32 nor float64.
TypeError: If `p` is not an int.
ValueError: If `p` is less than 1.
ValueError: If not all elements of `input` are greater than :math:`(p - 1) / 2`.
Supported Platforms:
``Ascend`` ``GPU`` ``CPU``
Examples:
>>> import mindspore
>>> import numpy as np
>>> from mindspore import Tensor, ops
>>> x = Tensor(np.array([[3, 4, 5], [4, 2, 6]]), mindspore.float32)
>>> y = ops.mvlgamma(x, p=3)
>>> print(y)
[[2.694925 5.402975 9.140645]
[5.402975 1.596312 13.64045]]
"""
mvlgamma_op = _get_cache_prim(Mvlgamma)(p)
return mvlgamma_op(input)
[docs]def nonzero(input, as_tuple=False):
r"""
Return the positions of all non-zero values.
Args:
input (Tensor): The input Tensor, its rank should be greater than or equal to 1.
as_tuple (bool, optional): Whether the output is tuple.
If ``False`` , return Tensor. Default: ``False`` .
If ``True`` , return Tuple of Tensor, only support ``Ascend`` .
Returns:
- If `as_tuple` is ``False``, return the Tensor, a 2-D Tensor whose data type is int64,
containing the positions of all non-zero values of the input.
- If `as_tuple` is ``True``, return the Tuple of Tensor and data type is int64.
The Tuple length is the dimension of the input tensor,
and each element is the 1D tensor of the subscript of all non-zero elements of
the input tensor in that dimension.
Raises:
TypeError: If `input` is not Tensor.
TypeError: If `as_tuple` is not bool.
ValueError: If dim of `input` equals to 0.
Supported Platforms:
``Ascend`` ``GPU`` ``CPU``
Examples:
>>> import mindspore
>>> import numpy as np
>>> from mindspore import Tensor, ops
>>> x = Tensor(np.array([[[1, 0], [-5, 0]]]), mindspore.int32)
>>> output = ops.nonzero(x)
>>> print(output)
[[0 0 0]
[0 1 0]]
>>> x = Tensor(np.array([1, 0, 2, 0, 3]), mindspore.int32)
>>> output = ops.nonzero(x, False)
>>> print(output)
[[0]
[2]
[4]]
>>> x = Tensor(np.array([[[1, 0], [-5, 0]]]), mindspore.int32)
>>> output = ops.nonzero(x, True)
>>> print(output)
(Tensor(shape=[2], dtype=Int64, value=[0, 0]),
Tensor(shape=[2], dtype=Int64, value=[0, 1]),
Tensor(shape=[2], dtype=Int64, value=[0, 0]))
>>> x = Tensor(np.array([1, 0, 2, 0, 3]), mindspore.int32)
>>> output = ops.nonzero(x, True)
>>> print(output)
(Tensor(shape=[3], dtype=Int64, value=[0, 2, 4]), )
"""
if as_tuple:
return non_zero_ext_(input)
return non_zero_(input)
[docs]def argwhere(input):
"""
Return a Tensor of the positions of all non-zero values.
Args:
input (Tensor): The input tensor. The data type is Number or Bool.
Returns:
Tensor, a 2-D Tensor whose data type is int64, containing the positions of all non-zero values of the input.
Raises:
TypeError: If `input` is not Tensor.
ValueError: If dim of `input` equals to 0.
Supported Platforms:
``Ascend`` ``GPU`` ``CPU``
Examples:
>>> import mindspore
>>> from mindspore import Tensor, ops
>>> import numpy as np
>>> x = Tensor(np.array([[[1, 0], [-5, 0]]]), mindspore.int32)
>>> output = ops.argwhere(x)
>>> print(output)
[[0 0 0]
[0 1 0]]
"""
return nonzero(input)
[docs]def column_stack(tensors):
"""
Stacks 1-D tensors as columns into a 2-D tensor. Tensors of other dimension are stacked as-is,
like :func:`mindspore.ops.hstack`.
Args:
tensors (Union[tuple[Tensor], list[Tensor]]): A sequence of tensors. All
of them must have the same shape except the axis to be concatenated.
Returns:
2-D Tensor, formed by stacking the given tensors.
Raises:
TypeError: If `tensors` is not list or tuple.
TypeError: If element in `tensors` is not Tensor.
ValueError: If `tensors` is empty.
Supported Platforms:
``Ascend`` ``GPU`` ``CPU``
Examples:
>>> from mindspore import Tensor, ops
>>> x1 = Tensor([1, 1, 1])
>>> x2 = Tensor([2, 2, 2])
>>> output = ops.column_stack((x1, x2))
>>> print(output)
[[1 2]
[1 2]
[1 2]]
"""
if not isinstance(tensors, (list, tuple)):
raise TypeError(
f"For column_stack, the input must be list or tuple of tensors, but got {type(tensors)}.")
trans_x = ()
for tensor in tensors:
if not isinstance(tensor, Tensor):
raise TypeError(
f"For column_stack, the input element must be tensor, but got {type(tensor)}.")
if tensor.ndim < 1:
tensor = expand_dims(tensor, 0)
if tensor.ndim == 1:
tensor = expand_dims(tensor, 1)
trans_x += (tensor,)
if not trans_x:
raise ValueError(
f"For column_stack, the input must have at least 1 tensor, but got 0.")
_concat = _get_cache_prim(P.Concat)(1)
return _concat(trans_x)
[docs]def hstack(tensors):
"""
Stacks tensors in sequence horizontally.
This is equivalent to concatenation along the second axis, except for 1-D tensors
where it concatenates along the first axis.
Args:
tensors (Union[tuple[Tensor], list[Tensor]]): A sequence of tensors. The
tensors must have the same shape along all but the second axis, except
1-D tensors which can be any length.
Returns:
Stacked Tensor, formed by stacking the given tensors.
Raises:
TypeError: If `tensors` is not list or tuple.
TypeError: If element in `tensors` is not Tensor.
ValueError: If `tensors` is empty.
Supported Platforms:
``Ascend`` ``GPU`` ``CPU``
Examples:
>>> from mindspore import Tensor, ops
>>> x1 = Tensor([1, 1, 1])
>>> x2 = Tensor([2, 2, 2])
>>> output = ops.hstack((x1, x2))
>>> print(output)
[1. 1. 1. 2. 2. 2.]
"""
if not isinstance(tensors, (list, tuple)):
raise TypeError(
f"For hstack, the input must be list or tuple, but got {type(tensors)}.")
tuple_of_tensor = ()
for tensor in tensors:
if not isinstance(tensor, Tensor):
raise TypeError(
f"For hstack, the input element must be tensor, but got {type(tensor)}.")
if tensor.ndim < 1:
tensor = expand_dims(tensor, 0)
tuple_of_tensor += (tensor,)
if not tuple_of_tensor:
raise ValueError(
"For hstack, the input must have at least 1 tensor, but got 0.")
if tuple_of_tensor[0].ndim <= 1:
_concat = _get_cache_prim(P.Concat)(0)
return _concat(tuple_of_tensor)
_concat = _get_cache_prim(P.Concat)(1)
return _concat(tuple_of_tensor)
@constexpr
def _check_axis_valid(axis, ndim):
"""
Checks axis are valid given ndim, and returns axis that can be passed
to the built-in operator (non-negative, int or tuple).
"""
if axis is None:
axis = ops.make_range(ndim)
return axis
if isinstance(axis, (tuple, list)):
axis = tuple(map(lambda x: _check_check_axis_in_range(x, ndim), axis))
return axis
return (_check_check_axis_in_range(axis, ndim),)
@constexpr
def _get_moved_perm(ndim, source, destination):
"""
Helper function for movedim, returns permutation after moving axis
from source to destination.
"""
dest_sorted_idx = [i for i, _ in sorted(
enumerate(destination), key=operator.itemgetter(1))]
axis_orig = [i for i in builtins.range(0, ndim) if i not in source]
k = 0
m = 0
perm = []
for i in dest_sorted_idx:
# inserts an axis that has been moved, denoted by n, and axis that remain
# in their original position, indexed from k to k + n - m, into index m in
# the list of permuted axis
n = destination[i]
j = k + n - m
perm += axis_orig[k:j]
perm.append(source[i])
k += n - m
m = n + 1
perm += axis_orig[k:]
return tuple(perm)
[docs]def movedim(x, source, destination):
"""
Moves axis of an array from source to destination.
Other axis remain in their original order.
Args:
x (Tensor): The tensor array whose axis should be reordered.
The dimension of `x` must not be 0.
source (Union[int, sequence[int]]): Original positions of the
axis to move. The length of `source` and `destination` must be the same.
destination (Union[int, sequence[int]]): Destination positions
for each of the original axis. The length of `source` and `destination` must be the same.
Returns:
Tensor, array with moved axis.
Raises:
ValueError: If axis are out of the range of `[-x.ndim, x.ndim)`, or
if the axis contain duplicates.
Supported Platforms:
``Ascend`` ``GPU`` ``CPU``
Examples:
>>> # case1 : moving single axis
>>> from mindspore import ops, Tensor
>>> import numpy as np
>>> x = Tensor(np.zeros((3, 4, 5)))
>>> output = ops.movedim(x, 0, -1)
>>> print(output.shape)
(4, 5, 3)
>>> # case 2 : moving multiple axes
>>> from mindspore import ops, Tensor
>>> import numpy as np
>>> x = Tensor(np.zeros((3, 4, 5)))
>>> output = ops.movedim(x, (0, 2), (1, 2))
>>> print(output.shape)
(4, 3, 5)
"""
ndim = ops.rank(x)
source = _check_axis_valid(source, ndim)
destination = _check_axis_valid(destination, ndim)
if len(source) != len(destination):
raise ValueError(
f"For `source` and `destination` arguments, the number of elements must be the same, but got 'source':"
f" {len(source)} and 'destination': {len(destination)}.")
perm = _get_moved_perm(ndim, source, destination)
return transpose_(x, perm)
[docs]def moveaxis(x, source, destination):
"""
Alias for `ops.movedim`. Moves axis of an array from source to destination.
Refer to :func:`mindspore.ops.movedim` for more detail.
Supported Platforms:
``Ascend`` ``GPU`` ``CPU``
Examples:
>>> from mindspore import ops, Tensor
>>> import numpy as np
>>> x = Tensor(np.zeros((3, 4, 5)))
>>> output = ops.moveaxis(x, 0, -1)
>>> print(output.shape)
(4, 5, 3)
"""
return movedim(x, source, destination)
@_primexpr
def _check_swapaxes_axis(axes, ndim):
return validator.check_swapaxes_axis(axes, ndim)
[docs]def swapaxes(input, axis0, axis1):
'''
Interchange two axes of a tensor.
Args:
input(Tensor): Input tensor.
axis0 (int): First axis.
axis1 (int): Second axis.
Returns:
Transposed tensor, has the same data type as `input`.
Raises:
TypeError: If argument `input` is not Tensor.
TypeError: If `axis0` or `axis1` is not integer.
ValueError: If `axis0` or `axis1` is not in the range of :math:`[-ndim, ndim-1]`.
Supported Platforms:
``Ascend`` ``GPU`` ``CPU``
Examples:
>>> import numpy as np
>>> from mindspore import ops
>>> from mindspore import Tensor
>>> input = Tensor(np.ones((2,3,4), dtype=np.float32))
>>> output = ops.swapaxes(input, 0, 2)
>>> print(output.shape)
(4, 3, 2)
'''
if not isinstance(input, Tensor):
raise TypeError(
f'For ops.swapaxes, parameter `input` must be Tensor, but got {type(input)}')
axis0, axis1 = _check_swapaxes_axis((axis0, axis1), input.ndim)
if axis0 == axis1:
return input
if axis0 > axis1:
axis0, axis1 = axis1, axis0
perm = ops.make_range(0, input.ndim)
if axis1 + 1 < input.ndim:
new_perm = perm[0:axis0] + perm[axis1:axis1 + 1] + \
perm[axis0 + 1:axis1] + perm[axis0:axis0 + 1] + perm[axis1 + 1:]
else:
new_perm = perm[0:axis0] + perm[axis1:axis1 + 1] + \
perm[axis0 + 1:axis1] + perm[axis0:axis0 + 1]
return transpose_(input, new_perm)
[docs]def swapdims(input, dim0, dim1):
'''
Interchange two dims of a tensor.
This function is equivalent to :func:`mindspore.ops.swapaxes` function.
Args:
input(Tensor): Input tensor.
dim0 (int): First dim.
dim1 (int): Second dim.
Returns:
Transposed tensor, has the same data type as `input`.
Raises:
TypeError: If argument `input` is not Tensor.
TypeError: If `dim0` or `dim1` is not integer.
ValueError: If `dim0` or `dim1` is not in the range of :math:`[-ndim, ndim-1]`.
Supported Platforms:
``Ascend`` ``GPU`` ``CPU``
Examples:
>>> import numpy as np
>>> from mindspore import ops
>>> from mindspore import Tensor
>>> input = Tensor(np.ones((2,3,4), dtype=np.float32))
>>> output = ops.swapdims(input, 0, 2)
>>> print(output.shape)
(4, 3, 2)
'''
return ops.swapaxes(input, dim0, dim1)
@constexpr
def _check_is_int(arg_value, arg_name, op_name):
arg_value = validator.check_is_int(arg_value, arg_name, op_name)
return arg_value
@_primexpr
def _check_positive_int(arg_value, arg_name, op_name):
arg_value = validator.check_int_range(
arg_value, 0, 2147483647, validator.INC_RIGHT, arg_name, op_name)
return arg_value
@constexpr
def _check_axis_range(arg_value, limit, arg_name, op_name):
arg_value = validator.check_int_range(
arg_value, -limit, limit, validator.INC_LEFT, arg_name, op_name)
return arg_value
@_primexpr
def _cal_repeat_dims(x_rank, rep, expand_axis):
rep_dims = [1] * (x_rank + 1)
rep_dims[expand_axis] = rep
return tuple(rep_dims)
@_primexpr
def _cal_reshape(x_shape, rep, axis):
x_reshape = list(x_shape)
x_reshape[axis] *= rep
return tuple(x_reshape)
@_primexpr
def _check_rank_range(x_rank, limit, arg_name, op_name):
if x_rank > limit:
raise ValueError(
f"For {op_name}, the rank of {arg_name} should be less than or equal to {limit}, but got {x_rank}.")
return x_rank
[docs]def repeat_interleave(input, repeats, axis=None):
"""
Repeat elements of a tensor along an axis, like `numpy.repeat`.
Args:
input (Tensor): The tensor to repeat values for. Must be of type: float16,
float32, int8, uint8, int16, int32, or int64.
repeats (Union[int, tuple, list, Tensor]): The number of times to repeat, must be positive.
axis (int, optional): The axis along which to repeat, Default: ``None``. if dims is None,
the input Tensor will be flattened and the output will alse be flattened.
Returns:
One tensor with values repeated along the specified axis. If input has shape
:math:`(s1, s2, ..., sn)` and axis is i, the output will have shape :math:`(s1, s2, ...,
si * repeats, ..., sn)`. The output type will be the same as the type of `input`.
Supported Platforms:
``Ascend`` ``GPU`` ``CPU``
Examples:
>>> import mindspore
>>> import numpy as np
>>> from mindspore import Tensor, ops
>>> input = Tensor(np.array([[0, 1, 2], [3, 4, 5]]), mindspore.int32)
>>> output = ops.repeat_interleave(input, repeats=2, axis=0)
>>> print(output)
[[0 1 2]
[0 1 2]
[3 4 5]
[3 4 5]]
"""
if axis is None:
input = input.reshape(-1)
axis = 0
if isinstance(repeats, Tensor):
repeats = TensorToList()(repeats)
output = input.repeat(repeats, axis)
return output
[docs]def repeat_interleave_ext(input, repeats, dim=None, output_size=None):
r"""
Repeat elements of a tensor along an axis, like `numpy.repeat`.
.. warning::
Only support on Atlas A2 training series.
Args:
input (Tensor): The tensor to repeat values for. Must be of type: float16,
float32, int8, uint8, int16, int32, or int64.
repeats (Union[int, tuple, list, Tensor]): The number of times to repeat, must be positive.
dim (int, optional): The dim along which to repeat, Default: ``None``. if dims is None,
the input Tensor will be flattened and the output will alse be flattened.
output_size (int, optional): Total output size for the given axis (e.g. sum of repeats),
Default: ``None``.
Returns:
One tensor with values repeated along the specified dim. If input has shape
:math:`(s1, s2, ..., sn)` and dim is i, the output will have shape :math:`(s1, s2, ...,
si * repeats, ..., sn)`. The output type will be the same as the type of `input`.
Supported Platforms:
``Ascend``
Examples:
>>> import mindspore
>>> import numpy as np
>>> from mindspore import Tensor, ops
>>> input = Tensor(np.array([[0, 1, 2], [3, 4, 5]]), mindspore.int32)
>>> output = ops.function.array_func.repeat_interleave_ext(input, repeats=2, dim=0)
>>> print(output)
[[0 1 2]
[0 1 2]
[3 4 5]
[3 4 5]]
"""
if isinstance(repeats, int):
return repeat_interleave_int_(input, repeats, dim, output_size)
return repeat_interleave_tensor_(input, repeats, dim, output_size)
[docs]def repeat_elements(x, rep, axis=0):
"""
Repeat elements of a tensor along an axis, like `numpy.repeat` .
Note:
It is recommended to use :func:'mindspore.mint.repeat_interleave', the dimension of input 'x' can support
a maximum of 8, and get better performance.
Args:
x (Tensor): The tensor to repeat values for. Must be of type: float16, float32, int8, uint8, int16, int32,
or int64. The rank of `x` must be less than or equal to 7.
rep (int): The number of times to repeat, must be positive.
axis (int): The axis along which to repeat. Default: 0.
Returns:
One tensor with values repeated along the specified axis. If x has shape
:math:`(s1, s2, ..., sn)` and axis is i, the output will have shape :math:`(s1, s2, ..., si * rep, ..., sn)`.
The output type will be the same as the type of `x`.
Raises:
ValueError: If the rank of `x` is greater than 7.
Supported Platforms:
``Ascend`` ``GPU`` ``CPU``
Examples:
>>> import mindspore
>>> import numpy as np
>>> from mindspore import Tensor, ops
>>> # case 1 : repeat on axis 0
>>> x = Tensor(np.array([[0, 1, 2], [3, 4, 5]]), mindspore.int32)
>>> output = ops.repeat_elements(x, rep = 2, axis = 0)
>>> print(output)
[[0 1 2]
[0 1 2]
[3 4 5]
[3 4 5]]
>>> # case 2 : repeat on axis 1
>>> x = Tensor(np.array([[0, 1, 2], [3, 4, 5]]), mindspore.int32)
>>> output = ops.repeat_elements(x, rep = 2, axis = 1)
>>> print(output)
[[0 0 1 1 2 2]
[3 3 4 4 5 5]]
"""
const_utils.check_type_valid(ops.dtype(x), mstype.number_type, 'input x')
rep = _check_positive_int(rep, "rep", "repeat_elements")
axis = _check_is_int(axis, "axis", "repeat_elements")
x_rank = rank_(x)
x_rank = _check_rank_range(x_rank, 7, "x", "repeat_elements")
axis = _check_axis_range(axis, x_rank, "axis", "repeat_elements")
axis = axis + x.ndim if axis < 0 else axis
expand_axis = axis + 1
x_expand = expand_dims(x, expand_axis)
rep_dims = _cal_repeat_dims(x_rank, rep, expand_axis)
x_expand = tile_(x_expand, rep_dims)
x_shape = shape_(x)
x_reshape = _cal_reshape(x_shape, rep, axis)
x_rep = reshape_(x_expand, x_reshape)
return x_rep
[docs]def sequence_mask(lengths, maxlen=None):
"""
Returns a mask tensor representing the first N positions of each cell.
If `lengths` has shape :math:`(d_1, d_2, ..., d_n)`, then the resulting tensor mask has type and shape
:math:`(d_1, d_2, ..., d_n, maxlen)`, with mask :math:`[i_1, i_2, ..., i_n, j] = (j < lengths[i_1, i_2, ..., i_n])`.
Args:
lengths (Tensor): Tensor to calculate the mask for. All values in this tensor should be
less than or equal to `maxlen`. Values greater than `maxlen` will be treated as `maxlen`.
maxlen (int): size of the last dimension of returned tensor. Must be positive and same
type as elements in `lengths`. Default is ``None`` .
Returns:
One mask tensor of shape `lengths.shape + (maxlen,)` .
Raises:
TypeError: If `lengths` is not a Tensor.
TypeError: If `maxlen` is not an int.
TypeError: If dtype of `lengths` is neither int32 nor int64.
Supported Platforms:
``GPU`` ``CPU``
Examples:
>>> import numpy as np
>>> from mindspore import Tensor, ops
>>> # case 1: When maxlen is assigned
>>> x = Tensor(np.array([1, 2, 3, 4]))
>>> output = ops.sequence_mask(x, 5)
>>> print(output)
[[ True False False False False]
[ True True False False False]
[ True True True False False]
[ True True True True False]]
>>> # case 2: When there is 0 in x
>>> x = Tensor(np.array([[1, 3], [2, 0]]))
>>> output = ops.sequence_mask(x, 5)
>>> print(output)
[[[ True False False False False]
[ True True True False False]]
[[ True True False False False]
[False False False False False]]]
>>> # case 3: when the maxlen is not assigned
>>> x = Tensor(np.array([[1, 3], [2, 4]]))
>>> output = ops.sequence_mask(x)
>>> print(output)
[[[ True False False False ]
[ True True True False ]]
[[ True True False False ]
[ True True True True ]]]
"""
const_utils.check_type_valid(
ops.dtype(lengths), [mstype.int64, mstype.int32], 'lengths')
if maxlen is None:
flatten_data = reshape_(lengths, (-1,))
flatten_data = cast_(flatten_data, mstype.float32)
_, value = arg_max_with_value_(flatten_data)
maxlen = cast_(value, mstype.int32)
else:
maxlen = _check_positive_int(maxlen, "maxlen", "sequence_mask")
maxlen = scalar_to_tensor_(maxlen, mstype.int32)
range_vector = range_(scalar_to_tensor_(0, mstype.int32),
maxlen, scalar_to_tensor_(1, mstype.int32))
mask = expand_dims(lengths, -1)
result = range_vector < mask
return result
def top_k(input_x, k, sorted=True):
r"""
`top_k` is deprecated, please use `ops.topk` instead.
"""
top_k_ = _get_cache_prim(P.TopK)(sorted)
return top_k_(input_x, k)
[docs]def gather_ext(input, dim, index):
r"""
Gather data from a tensor by indices.
.. math::
output[(i_0, i_1, ..., i_{dim}, i_{dim+1}, ..., i_n)] =
input[(i_0, i_1, ..., index[(i_0, i_1, ..., i_{dim}, i_{dim+1}, ..., i_n)], i_{dim+1}, ..., i_n)]
.. warning::
On Ascend, the behavior is unpredictable in the following cases:
- the value of `index` is not in the range `[-input.shape[dim], input.shape[dim])` in forward;
- the value of `index` is not in the range `[0, input.shape[dim])` in backward.
Args:
input (Tensor): The target tensor to gather values.
dim (int): the axis to index along, must be in range `[-input.rank, input.rank)`.
index (Tensor): The index tensor, with int32 or int64 data type. An valid `index` should be:
- `index.rank == input.rank`;
- for `axis != dim`, `index.shape[axis] <= input.shape[axis]`;
- the value of `index` is in range `[-input.shape[dim], input.shape[dim])`.
Returns:
Tensor, has the same type as `input` and the same shape as `index`.
Raises:
ValueError: If the shape of `index` is illegal.
ValueError: If `dim` is not in `[-input.rank, input.rank)`.
ValueError: If the value of `index` is out of the valid range.
TypeError: If the type of `index` is illegal.
Supported Platforms:
``Ascend`` ``GPU`` ``CPU``
Examples:
>>> import mindspore
>>> import numpy as np
>>> from mindspore import Tensor, ops
>>> from mindspore.ops.function.array_func import gather_ext
>>> input = Tensor(np.array([[-0.1, 0.3, 3.6], [0.4, 0.5, -3.2]]), mindspore.float32)
>>> index = Tensor(np.array([[0, 0], [1, 1]]), mindspore.int32)
>>> output = gather_ext(input, 1, index)
>>> print(output)
[[-0.1 -0.1]
[0.5 0.5]]
"""
return gather_d_op(input, dim, index)
[docs]def max_ext(input, dim=None, keepdim=False):
"""
Calculates the maximum value along with the given dimension for the input tensor.
Args:
input (Tensor): The input tensor, can be any dimension. Complex tensor is not supported for now.
dim (int, optional): The dimension to reduce. Default: ``None`` .
keepdim (bool, optional): Whether to reduce dimension, if true, the output will keep same dimension
with the input, the output will reduce dimension if false. Default: ``False`` .
Returns:
Tensor if `dim` is the default value ``None`` , the maximum value of input tensor, with the shape :math:`()` ,
and same dtype as `input`.
tuple (Tensor) if `dim` is not the default value ``None`` , tuple of 2 tensors, containing the maximum
value of the input tensor along the given dimension `dim` and the corresponding index.
- **values (Tensor)** - The maximum value of input tensor along the given dimension `dim`, with same dtype as
`input`. If `keepdim` is ``True`` , the shape of output tensors is :math:`(input_1, input_2, ...,
input_{axis-1}, 1, input_{axis+1}, ..., input_N)` . Otherwise, the shape is :math:`(input_1, input_2, ...,
input_{axis-1}, input_{axis+1}, ..., input_N)` .
- **index (Tensor)** - The index for the maximum value of the input tensor along the given dimension `dim`, with
the same shape as `values`.
Raises:
ValueError: If `dim` is the default value ``None`` and `keepdim` is not ``False`` .
Supported Platforms:
``Ascend`` ``GPU`` ``CPU``
Examples:
>>> import mindspore
>>> import numpy as np
>>> from mindspore import Tensor, ops
>>> from mindspore.ops.function.array_func import max_ext
>>> y = Tensor(np.array([[0.0, 0.3, 0.4, 0.5, 0.1],
... [3.2, 0.4, 0.1, 2.9, 4.0]]), mindspore.float32)
>>> output, index = max_ext(y, 0, True)
>>> print(output, index)
[[3.2 0.4 0.4 2.9 4. ]] [[1 1 0 1 1]]
"""
if dim is None:
if keepdim is not False:
raise ValueError(
f"For 'max', the `keepdim` must be False when the `dim` is None, but got {keepdim}")
return max_(input)
argmax_with_value_op = _get_cache_prim(ArgMaxWithValue)(dim, keepdim)
indices, values = argmax_with_value_op(input)
return values, indices
[docs]def min_ext(input, dim=None, keepdim=False):
"""
Calculates the minimum value along with the given dimension for the input tensor.
Args:
input (Tensor): The input tensor, can be any dimension. Complex tensor is not supported for now.
dim (int, optional): The dimension to reduce. Default: ``None`` .
keepdim (bool, optional): Whether to reduce dimension, if true, the output will keep same dimension
with the input, the output will reduce dimension if false. Default: ``False`` .
Returns:
Tensor if `dim` is the default value ``None`` , the minimum value of input tensor, with the shape :math:`()` ,
and same dtype as `input`.
tuple (Tensor) if `dim` is not the default value ``None`` , tuple of 2 tensors, containing the minimum value
of the input tensor along the given dimension `dim` and the corresponding index.
- **values (Tensor)** - The minimum value of input tensor along the given dimension `dim`, with same dtype as
`input`. If `keepdim` is ``True`` , the shape of output tensors is :math:`(input_1, input_2, ...,
input_{axis-1}, 1, input_{axis+1}, ..., input_N)` . Otherwise, the shape is :math:`(input_1, input_2, ...,
input_{axis-1}, input_{axis+1}, ..., input_N)` .
- **index (Tensor)** - The index for the minimum value of the input tensor along the given dimension `dim`,
with the same shape as `values`.
Raises:
ValueError: If `dim` is the default value ``None`` and `keepdim` is not ``False`` .
Supported Platforms:
``Ascend`` ``GPU`` ``CPU``
Examples:
>>> import mindspore
>>> import numpy as np
>>> from mindspore import Tensor, ops
>>> from mindspore.ops.function.array_func import min_ext
>>> x = Tensor(np.array([0.0, 0.4, 0.6, 0.7, 0.1]), mindspore.float32)
>>> output, index = min_ext(x, 0, keepdim=True)
>>> print(output, index)
[0.0] [0]
"""
if dim is None:
if keepdim is not False:
raise ValueError(
f"For 'min', the `keepdim` must be False when the `dim` is None, but got {keepdim}")
return min_(input)
argmin_with_value_op = _get_cache_prim(ArgMinWithValue)(dim, keepdim)
indices, values = argmin_with_value_op(input)
return values, indices
def one_hot_ext(tensor, num_classes):
r"""
Computes a one-hot tensor.
The locations represented by tensor in `tensor` take value `1`, while all
other locations take value `0`.
Args:
tensor (Tensor): A tensor of indices. Tensor of shape :math:`(X_0, \ldots, X_n)`.
Data type must be int32 or int64.
num_classes (int): A scalar defining the depth of the one-hot dimension.
Returns:
Tensor, one-hot tensor.
Raises:
TypeError: If `num_classes` is not an int.
TypeError: If dtype of `tensor` is not int32 or int64.
ValueError: If `num_classes` is less than 0.
Supported Platforms:
``Ascend`` ``GPU`` ``CPU``
Examples:
>>> import mindspore
>>> import numpy as np
>>> from mindspore import ops
>>> from mindspore import Tensor
>>> from mindspore.ops.function.array_func import one_hot_ext
>>> tensor = Tensor(np.array([0, 1, 2]), mindspore.int32)
>>> num_classes = 3
>>> output = one_hot_ext(tensor, num_classes)
>>> print(output)
[[1. 0. 0.]
[0. 1. 0.]
[0. 0. 1.]]
"""
on_value = Tensor(1, dtype=tensor.dtype)
off_value = Tensor(0, dtype=tensor.dtype)
return one_hot_ext_impl(tensor, num_classes, on_value, off_value, -1)
[docs]def from_numpy(array):
r"""
Convert numpy array to Tensor.
If the data is not C contiguous, the data will be copied to C contiguous to construct the tensor.
Otherwise, the tensor will be constructed using this numpy array without copy.
Args:
array (numpy.array): The input array.
Returns:
Tensor, has the same data type as input array.
Examples:
>>> import numpy as np
>>> import mindspore as ms
>>> x = np.array([1, 2])
>>> output = ms.from_numpy(x)
>>> print(output)
[1 2]
"""
return Tensor.from_numpy(array)
__all__ = [
'unique',
'unique_with_pad',
'unique_consecutive',
'eye',
'matrix_band_part',
'padding',
'fill',
'fills',
'tile',
'size',
'ger',
'ones',
'ones_like',
'zeros',
'zeros_like',
'zero_',
'shape',
'shape_',
'reverse',
'reverse_sequence',
'hamming_window',
'chunk',
'full',
'full_like',
'dyn_shape',
'rank',
'arange',
'range',
'reshape',
'reshape_',
'flatten',
'tensor_slice',
'strided_slice',
'slice',
'slice_scatter',
'select_scatter',
'cat',
'concat',
'stack',
'unbind',
'unstack',
'is_tensor',
'scalar_cast',
'scalar_to_array',
'scalar_to_tensor',
'space_to_batch_nd',
'batch_to_space_nd',
'tuple_to_array',
'expand_dims',
'squeeze',
'unsqueeze',
'transpose',
'scatter_nd',
'scatter_nd_add',
'scatter_nd_sub',
'scatter_nd_mul',
'scatter_nd_div',
'scatter_nd_max',
'scatter_nd_min',
'tensor_scatter_add',
'tensor_scatter_sub',
'tensor_scatter_mul',
'tensor_scatter_div',
'tensor_scatter_max',
'tensor_scatter_min',
'tensor_scatter_elements',
'scatter',
'unsorted_segment_min',
'unsorted_segment_max',
'unsorted_segment_prod',
'gather',
'gather_d',
'gather_elements',
'gather_nd',
'one_hot',
'masked_fill',
'masked_select',
'where',
'narrow',
'ravel',
'scatter_add',
'scatter_mul',
'scatter_max',
'scatter_min',
'scatter_div',
'scatter_update',
'select',
'tril',
'triu',
'nonzero',
'is_nonzero',
'matrix_diag',
'matrix_diag_part',
'matrix_set_diag',
'diag',
'diagflat',
'meshgrid',
'affine_grid',
'meshgrid',
'broadcast_to',
'col2im',
'split',
'tensor_split',
'vsplit',
'hsplit',
'dsplit',
'index_fill',
'index_select',
'max',
'argmax',
'min',
'unsorted_segment_sum',
'population_count',
'topk',
'expand',
'fold',
'unfold',
'diagonal',
'diagonal_scatter',
'lstsq',
'mvlgamma',
'swapaxes',
'swapdims',
'searchsorted',
'argsort',
'sequence_mask',
'repeat_elements',
'repeat_interleave',
'argwhere',
'column_stack',
'hstack',
'movedim',
'moveaxis',
'aminmax',
'sort',
'top_k',
'deepcopy',
'flip',
]
__all__.sort()