mars.tensor.compress#

mars.tensor.compress(condition, a, axis=None, out=None)[源代码]#

Return selected slices of a tensor along given axis.

When working along a given axis, a slice along that axis is returned in output for each index where condition evaluates to True. When working on a 1-D array, compress is equivalent to extract.

参数

  • condition (1-D tensor of bools) – Tensor that selects which entries to return. If len(condition) is less than the size of a along the given axis, then output is truncated to the length of the condition tensor.

  • a (array_like) – Tensor from which to extract a part.

  • axis (int, optional) – Axis along which to take slices. If None (default), work on the flattened tensor.

  • out (Tensor, optional) – Output tensor. Its type is preserved and it must be of the right shape to hold the output.

返回

compressed_array – A copy of a without the slices along axis for which condition is false.

返回类型

Tensor

参见

take, choose, diag, diagonal, select

Tensor.compress

Equivalent method in ndarray

mt.extract

Equivalent method when working on 1-D arrays

示例

>>> import mars.tensor as mt

>>> a = mt.array([[1, 2], [3, 4], [5, 6]])
>>> a.execute()
array([[1, 2],
       [3, 4],
       [5, 6]])
>>> mt.compress([0, 1], a, axis=0).execute()
array([[3, 4]])
>>> mt.compress([False, True, True], a, axis=0).execute()
array([[3, 4],
       [5, 6]])
>>> mt.compress([False, True], a, axis=1).execute()
array([[2],
       [4],
       [6]])

Working on the flattened tensor does not return slices along an axis but selects elements.

>>> mt.compress([False, True], a).execute()
array([2])