flux#

class diffsptk.Flux(lag=1, norm=2, reduction='mean')[source]#

Flux calculation module.

Parameters:
lagint

Lag of the distance calculation, \(L\).

normint or float

Order of norm.

reduction[‘none’, ‘mean’, ‘batchmean’, ‘sum’]

Reduction type.

forward(x, y=None)[source]#

Calculate flux, which is the distance between adjacent frames.

Parameters:
xTensor [shape=(…, N, D)]

Input.

yTensor [shape=(…, N, D)] or None

Target (optional).

Returns:
outTensor [shape=(…, N-|L|) or scalar]

Flux.

Examples

>>> x = diffsptk.ramp(5).view(3, 2)
>>> x
tensor([[0., 1.],
        [2., 3.],
        [4., 5.]])
>>> flux = diffsptk.Flux(reduction="none", norm=1)
>>> f = flux(x)
>>> f
tensor([4., 4.])
diffsptk.functional.flux(x, y=None, lag=1, norm=2, reduction='mean')[source]#

Calculate flux.

Parameters:
xTensor [shape=(…, N, D)]

Input.

yTensor [shape=(…, N, D)] or None

Target (optional).

lagint

Lag of the distance calculation, \(L\).

normint or float

Order of norm.

reduction[‘none’, ‘mean’, ‘batchmean’, ‘sum’]

Reduction type.

Returns:
outTensor [shape=(…, N-|L|) or scalar]

Flux.

See also

rmse