lucid.random.bernoulli

lucid.random.bernoulli(probs: int | float | complex | list[int | float | complex] | ndarray | array | Tensor, requires_grad: bool = False, keep_grad: bool = False, device: Literal['cpu', 'gpu'] = 'cpu') → Tensor

The bernoulli function generates a tensor by sampling from a Bernoulli distribution for each element in the given probability tensor. Each element in the output tensor is either 0 or 1 based on the corresponding probability value.

Function Signature

def bernoulli(
    probs: _ArrayOrScalar | Tensor,
    requires_grad: bool = False,
    keep_grad: bool = False,
    device: _DeviceType = "cpu",
) -> Tensor

Parameters

  • probs (_ArrayOrScalar | Tensor): A tensor or scalar representing probabilities for each element to be 1. Values must lie in the range [0, 1].

  • requires_grad (bool, optional): If True, the resulting tensor will track gradients for automatic differentiation. Defaults to False.

  • keep_grad (bool, optional): Determines whether gradient history should persist across multiple operations. Defaults to False.

Returns

  • Tensor: A tensor of the same shape as probs, with elements sampled from a Bernoulli distribution (0 or 1).

Example

>>> import lucid
>>> x = lucid.random.bernoulli([0.2, 0.8, 0.5])
>>> print(x)
Tensor([0, 1, 0], grad=None)

By default, the generated tensor does not track gradients. To enable gradient tracking, set requires_grad=True:

>>> probs = lucid.Tensor([0.3, 0.7, 0.9])
>>> y = lucid.random.bernoulli(probs, requires_grad=True)
>>> print(y.requires_grad)
True

Note

  • The probs values must be in the range [0, 1].

  • Use lucid.random.seed to ensure reproducibility of random samples.

  • Since this function involves discrete random sampling, gradients cannot be propagated through it.