biomedical_data_generator.utils.sampling

Utility functions for sampling from distributions.

Functions

sample_distribution(distribution, params, ...)

Draw a block of independent samples from a NumPy RNG distribution.

biomedical_data_generator.utils.sampling.sample_distribution(distribution, params, rng, size)

Draw a block of independent samples from a NumPy RNG distribution.

The distribution string is expected to match the corresponding numpy.random.Generator method name, except for “exp_normal”, which is implemented as exp(rng.normal(…)).

All parameter names and values are validated in DatasetConfig / ClassConfig via validate_distribution_params, so this function only dispatches to the correct RNG method. None params resolve to an empty mapping, falling back to the RNG method’s own defaults.

Parameters:

• distribution (Literal['normal', 'lognormal', 'exp_normal', 'uniform', 'exponential', 'laplace'])

• params (dict[str, Any] | None)

• rng (Generator)

• size (tuple[int, int])

Return type:

ndarray