randint

deeptrack.backend.array_api_compat_ext.torch.random.randint(low: int, high: int | None = None, size: tuple[int, ...] | None = None) → Tensor

Sample integers from a uniform discrete distribution.

Mirrors numpy.random.randint.

Parameters

low: int

Lowest integer (inclusive) if high is provided. If high is None, this is treated as the exclusive upper bound, and low is set to 0.

high: int | None, optional

Upper bound (exclusive).

size: tuple[int, …] | None, optional

Output shape. If None, returns a scalar tensor.

Returns

torch.Tensor

Random integers in [low, high) with dtype torch.int64.

Examples

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

>>> rnd.randint(5)
tensor(3)

>>> rnd.randint(2, 10, (2, 3)).shape
torch.Size([2, 3])