_containers#

ns_sdn.engine._core._containers = <module 'ns_sdn.engine._core._containers' from '/home/runner/work/ns-sdn/ns-sdn/src/ns_sdn/engine/_core/_containers.py'>[source]#

Reusable value objects shared by internal mechanisms and public results.

Five records, none of which knows what a model is: the penalty weights and the training stage that make up a schedule, the information-criteria triple, the affine scaling map, and the summary table. Each is a frozen dataclass or a named tuple, validated at construction where it carries numbers, and importable from here by the engine while the public modules re-bind the user-facing ones (Penalties, Stage, InformationCriteria, SummaryTable) under their public names.

Example

>>> stage = _Stage(1e-3, 10, _Penalties(amplitude=0.0))
>>> stage.penalties.frequency, _MinMaxScale.fit(np.array([0.0, 4.0])).slope
(0.0008, 2.0)