lucid.reciprocal

lucid.reciprocal(a: Tensor, /) Tensor

The reciprocal function computes the element-wise reciprocal of each element in the input tensor.

Function Signature

def reciprocal(a: Tensor) -> Tensor

Parameters

  • a (Tensor): The input tensor for which reciprocals are computed.

Returns

  • Tensor:

    A new tensor containing the reciprocals of the elements in the input tensor. If a requires gradients, the resulting tensor will also require gradients.

Forward Calculation

\[\mathbf{out}_i = \frac{1}{\mathbf{a}_i}\]

Backward Gradient Calculation

\[\frac{\partial \mathbf{out}_i}{\partial \mathbf{a}_i} = -\frac{1}{\mathbf{a}_i^2}\]

Example

>>> import lucid
>>> a = Tensor([1, 2, 4], requires_grad=True)
>>> out = lucid.reciprocal(a)
>>> print(out)
Tensor([1. 0.5 0.25], grad=None)