compute_ser

sionna.phy.utils.compute_ser(s: torch.Tensor, s_hat: torch.Tensor, precision: Literal['single', 'double'] = 'double') → torch.Tensor

Computes the symbol error rate (SER) between two integer tensors.

Parameters:

• s (torch.Tensor) – A tensor of arbitrary shape filled with integers.

• s_hat (torch.Tensor) – A tensor like s.

• precision (Literal['single', 'double']) – Precision used for internal calculations and outputs. Defaults to "double".

Outputs:

ser – torch.float. SER.

Examples

import torch
from sionna.phy.utils import compute_ser

s = torch.tensor([0, 1, 2, 3])
s_hat = torch.tensor([0, 1, 3, 2])
print(compute_ser(s, s_hat).item())
# 0.5