mindscience.sciops.fft.asd_irfftn

mindscience.sciops.fft.asd_irfftn(xr, xi, n=None, ndim=1)[source]

N-dimensional complex-to-real inverse FFT transform using Ascend NPU acceleration.

This function provides a unified interface for 1D and 2D complex-to-real inverse FFT transforms, optimized for Ascend NPU hardware acceleration.

Parameters

  • xr (Tensor) – Real part of input complex tensor with data type float32.

  • xi (Tensor) – Imaginary part of input complex tensor with data type float32.

  • n (int, optional) – Length of the output tensor. Default: None.

  • ndim (int, optional) – Number of dimensions to transform. Only support 1 and 2. Default: 1.

Returns

yr (Tensor). Output real tensor with data type float32.

Examples

>>> import mindspore as ms
>>> from mindscience.sciops.fft import asd_irfftn
>>> xr = ms.Tensor([[1.0, 2.0, 3.0]], ms.float32)
>>> xi = ms.Tensor([[0.0, 0.0, 0.0]], ms.float32)
>>> yr = asd_irfftn(xr, xi, ndim=1)
>>> print(yr.shape)
(1, 4)