mindspore.ops.scatter_nd_min

mindspore.ops.scatter_nd_min(input_x, indices, updates, use_locking=False)

Perform a sparse minimum update on input_x based on the specified indices and update values.

$$\text{input\_x}[\text{indices}[i,...,j]]=min(\text{input\_x}[\text{indices}[i,...,j]],\text{updates}[i,...,j])$$

Note

• Support implicit type conversion and type promotion.

• The dimension of indices is at least 2, and its shape must be indices.shape[-1] <= len(indices.shape).

• The shape of updates is indices.shape[:-1] + input_x.shape[indices.shape[-1]:].

Parameters

• input_x (Union[Parameter, Tensor]) – The input parameter or tensor.

• indices (Tensor) – The specified indices.

• updates (Tensor) – The update values.

• use_locking (bool) – Whether to protect the assignment by a lock. Default: False .

Returns

Tensor

Supported Platforms:

Ascend GPU CPU

Examples

>>> 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]]]