randn#

deeptrack.backend.array_api_compat_ext.torch.random.randn(*size: int) → Tensor#

Sample from the standard normal distribution.

This function mirrors numpy.random.randn, i.e. it takes the output shape as positional integer arguments.

Parameters#

*size: int: Output shape given as positional integers. If empty, returns a scalar 0D tensor.

torch.Tensor: A tensor of shape size (or scalar if size is empty) with values sampled from a standard normal distribution.

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

>>> rnd.randn(2, 3).shape
torch.Size([2, 3])

Scalar sample:

>>> rnd.randn()
tensor(-2.2435)