credit.ensemble.gaussian

Classes

GaussianNoise

Simple Gaussian white noise generator.

Module Contents

class credit.ensemble.gaussian.GaussianNoise(amplitude: float = 0.05)

Simple Gaussian white noise generator.

Generates uncorrelated Gaussian noise with zero mean and controllable standard deviation. Each point is independently sampled from a normal distribution.

Unlike spatially correlated noise (e.g., RedNoise), this produces completely independent random values at each grid point, which may be appropriate for: * Model parameter uncertainty * Observation errors * Simple perturbation schemes * Baseline comparisons with more sophisticated noise models

Parameters:

amplitude (float, optional) – Standard deviation of the Gaussian noise. Defaults to 0.05.

amplitude = 0.05

__call__(x: torch.Tensor) → torch.Tensor

Generate Gaussian white noise matching input tensor dimensions.

Parameters:

x (torch.Tensor) – Reference tensor whose shape determines the output noise dimensions.

Returns:

Gaussian white noise tensor with the same shape as input,

zero mean, and standard deviation equal to the amplitude parameter.

Return type:

torch.Tensor

__repr__() → str