cupyx.scipy.fftpack.get_fft_plan

cupyx.scipy.fftpack.get_fft_plan(a, shape=None, axes=None, value_type='C2C')

Generate a CUDA FFT plan for transforming up to three axes.

Parameters:
  • a (cupy.ndarray) – Array to be transform, assumed to be either C- or F- contiguous.
  • shape (None or tuple of ints) – Shape of the transformed axes of the output. If shape is not given, the lengths of the input along the axes specified by axes are used.
  • axes (None or int or tuple of int) –

    The axes of the array to transform. If None, it is assumed that all axes are transformed.

    Currently, for performing N-D transform these must be a set of up to three adjacent axes, and must include either the first or the last axis of the array.

  • value_type ('C2C') – The FFT type to perform. Currently only complex-to-complex transforms are supported.
Returns:

a cuFFT plan for either 1D transform (cupy.cuda.cufft.Plan1d)

or N-D transform (cupy.cuda.cufft.PlanNd).
Return type:

plan