pado.math.ifft

ifft(arr_c, normalized='backward', pad_width=None, shift=True)[source]

Compute 2D inverse FFT of a complex tensor with optional padding and shifting.

Parameters:

  • arr_c (torch.Tensor) – Complex tensor [B, Ch, H, W]

  • normalized (str) – IFFT normalization mode: “backward”, “forward”, or “ortho”

  • pad_width (tuple) – Padding as (left, right, top, bottom)

  • shift (bool) – If True, center zero-frequency component

Returns:

IFFT result tensor

Return type:

torch.Tensor

Examples

>>> field = torch.ones((1, 1, 64, 64), dtype=torch.complex64)
>>> field_freq = fft(field)
>>> field_restored = ifft(field_freq)