fft#

sionna.phy.signal.fft(tensor: torch.Tensor, axis: int = -1, precision: Literal['single', 'double'] | None = None) → torch.Tensor[source]#

Computes the normalized DFT along a specified axis.

This operation computes the normalized one-dimensional discrete Fourier transform (DFT) along the axis dimension of a tensor. For a vector \(\mathbf{x}\in\mathbb{C}^N\), the DFT \(\mathbf{X}\in\mathbb{C}^N\) is computed as

\[X_m = \frac{1}{\sqrt{N}}\sum_{n=0}^{N-1} x_n \exp \left\{ -j2\pi\frac{mn}{N}\right\},\quad m=0,\dots,N-1.\]
Parameters:

  • tensor (torch.Tensor) – Tensor of arbitrary shape (torch.complex)

  • axis (int) – Dimension along which the DFT is taken

  • precision (Literal['single', 'double'] | None) – Precision used for internal calculations and outputs. If set to None, precision is used.

Outputs:

x_f – torch.complex. Tensor of the same shape as tensor.

Examples

import torch
from sionna.phy.signal import fft

x = torch.randn(32, 64, dtype=torch.complex64)
X = fft(x)
print(X.shape)
# torch.Size([32, 64])
On this page

This Page