conv_partial

dnois.fourier.conv_partial(f1: Tensor, f2_ft: Tensor, f2_shape: Sequence[int], dims: Sequence[int], real: bool, out: Literal['full', 'same', 'valid'] = 'full', padding: int | Sequence[int] | str = 'linear', simpson: bool = True) → Tensor

See conv_mult(). DFT-scale is not applied here.

Parameters:

  • f1 (Tensor) – The first tensor for convolution.

  • f2_ft (Tensor) – The FFT of the second tensor for convolution. Typically produced by ft4conv().

  • f2_shape (Sequence[int]) – Shape of original second tensor. This is needed to ensure correct padding.

  • dims (Sequence[int]) – Dimensions to be convolved.

  • real (bool) – Whether to use FFT for real tensors to f1. This should be consistent with that for f2_ft.

  • out (str) – See dconv().

  • padding (int, Sequence[int] or str) – See dconv().

  • simpson (bool) – Whether to apply Simpson’s rule to f1 before FFT. Default: True.

Returns:

Convolution between f1 and f2 (whose FFT is f2_ft). See dconv() for descriptions about its dtype and shape.