mindspore.ops.scatter_nd_min

查看源文件
mindspore.ops.scatter_nd_min(input_x, indices, updates, use_locking=False)[源代码]

根据指定索引和更新值对 input_x 进行稀疏最小值更新。

input_x[indices[i,...,j]]=min(input_x[indices[i,...,j]],updates[i,...,j])

说明

  • 支持隐式类型转换、类型提升。

  • indices 的维度至少为2,并且 indices.shape[-1] <= len(indices.shape)

  • updates 的shape为 indices.shape[:-1] + input_x.shape[indices.shape[-1]:]

参数:
  • input_x (Union[Parameter, Tensor]) - 输入的parameter或tensor。

  • indices (Tensor) - 指定索引。

  • updates (Tensor) - 更新值。

  • use_locking (bool) - 是否启用锁保护。默认 False

返回:

Tensor

支持平台:

Ascend GPU CPU

样例:

>>> import mindspore
>>> import numpy as np
>>> input_x = mindspore.Parameter(mindspore.tensor(np.ones(8) * 10, mindspore.float32), name="x")
>>> indices = mindspore.tensor([[2], [4], [1], [7]], mindspore.int32)
>>> updates = mindspore.tensor([6, 7, 8, 9], mindspore.float32)
>>> output = mindspore.ops.scatter_nd_min(input_x, indices, updates, False)
>>> print(output)
[10.  8.  6. 10.  7. 10. 10.  9.]
>>> input_x = mindspore.Parameter(mindspore.tensor(np.ones((4, 4, 4)) * 10, mindspore.int32))
>>> indices = mindspore.tensor([[0], [2]], mindspore.int32)
>>> updates = mindspore.tensor([[[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 = mindspore.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]]]