permutation

deeptrack.backend.array_api_compat_ext.torch.random.permutation(x: int | Tensor) → Tensor

Return a permuted sequence or tensor.

Mirrors numpy.random.permutation.

Parameters

x: int | torch.Tensor

If an integer, returns a permutation of torch.arange(x). If a tensor, returns a permuted copy along the first axis.

Returns

torch.Tensor

A permuted tensor.

Examples

>>> import deeptrack.backend.array_api_compat_ext.torch.random as rnd

>>> rnd.permutation(5)
tensor([2, 4, 0, 1, 3])

>>> import torch
>>>
>>> a = torch.arange(12).reshape(3, 4)
>>> rnd.permutation(a).shape
torch.Size([3, 4])