AbstractNoiseModel

class pmrf.noise_models.AbstractNoiseModel

Bases: Module

Abstract base class for likelihood noise models.

A noise model maps a model prediction to a noise parameter, such as variance. For example, for a Gaussian likelihood, a noise model can be used to model the variance with non-standard broadcasting rules.

See pmrf.noise_models for built-in noise models.

abstractmethod __call__(y_event: Array) → Array | tuple[Array]

Map model predictions to noise parameters.

Parameters:

y_event (jnp.ndarray) – The mean model prediction in event space.

Returns:

The noise parameter or a tuple of noise parameters.

Return type:

jnp.ndarray | tuple[jnp.ndarray, jnp.ndarray]