ft4conv

dnois.fourier.ft4conv(f1_shape: Sequence[int], f2: Tensor, dims: Sequence[int], real: bool, spacing: float | Tensor | Sequence[float | Tensor] = None, padding: int | Sequence[int] | str = 'linear', simpson: bool = False) → Tensor

See conv_mult().

Parameters:

  • f1_shape (Sequence[int]) – Shape of the original first tensor for convolution. This is needed to ensure correct padding.

  • f2 (Tensor) – The second tensor for convolution.

  • dims (Sequence[int]) – Dimensions for convolution.

  • real (bool) – Whether to use FFT for real tensors to f2.

  • spacing – See conv().

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

  • simpson (bool) – Whether to apply Simpson’s rule to f2 before FFT. Default: True.

Returns: