Clamp

class pfhedge.nn.Clamp(*args, **kwargs)

Clamp all elements in input into the range \([\min, \max]\).

The bounds \(\min\) and \(\max\) can be tensors.

If \(\min \leq \max\):

\[\begin{split}\text{output} = \begin{cases} \min & \text{input} < \min \\ \text{input} & \min \leq \text{input} \leq \max \\ \max & \max < \text{input} \end{cases}\end{split}\]

If \(\min > \max\):

\[\begin{split}\text{output} = \begin{cases} \frac12 (\min + \max) & \text{inverted_output} = \text{'mean'} \\ \max & \text{inverted_output} = \text{'max'} \\ \end{cases}\end{split}\]

See also

• torch.clamp()

• pfhedge.nn.functional.clamp()

Parameters

inverted_output ({'mean', ''max'}, default='mean') – Controls the output when \(\min > \max\). ‘max’ is consistent with torch.clamp().

Shape:

• input: \((N, *)\) where \(*\) means any number of additional dimensions.

• min: \((N, *)\) or any size broadcastable to input.

• max: \((N, *)\) or any size broadcastable to input.

• output: \((N, *)\), same shape as the input.

Examples

>>> import torch
>>> from pfhedge.nn import Clamp
>>>
>>> m = Clamp()
>>> input = torch.linspace(-2, 12, 15) * 0.1
>>> input
tensor([-0.2000, -0.1000,  0.0000,  0.1000,  0.2000,  0.3000,  0.4000,  0.5000,
         0.6000,  0.7000,  0.8000,  0.9000,  1.0000,  1.1000,  1.2000])
>>> m(input, 0.0, 1.0)
tensor([0.0000, 0.0000, 0.0000, 0.1000, 0.2000, 0.3000, 0.4000, 0.5000, 0.6000,
        0.7000, 0.8000, 0.9000, 1.0000, 1.0000, 1.0000])

When \(\min > \max\), returns the mean of \(\min\) and \(\max\).

>>> input = torch.tensor([1.0, 0.0])
>>> min = torch.tensor([0.0, 1.0])
>>> max = torch.tensor([0.0, 0.0])
>>> m(input, min, max)
tensor([0.0000, 0.5000])

forward(input, min=None, max=None)

Clamp all elements in input into the range \([\min, \max]\).

Parameters

• input (torch.Tensor) – The input tensor.

• min (torch.Tensor, optional) – Lower-bound of the range to be clamped to.

• max (torch.Tensor, optional) – Upper-bound of the range to be clamped to.

Returns

torch.Tensor