hydromodpy.calibration.optimizer

Optimizer Protocol and registry.

An optimizer proposes parameter points (ask) and ingests evaluation results (tell). Adapters for scipy, optuna, grid-search are found under hydromodpy/calibration/adapters/ and registered here.

Module attributes

FAILED_EVAL_COST

Sentinel cost used when an evaluation fails or returns NaN.

Functions

available_optimizers()	Return registered optimizer names, including installed entry points.
build_optimizer(name, space, **kwargs)	Construct an optimizer by name.
register_optimizer(name)	Register a built-in optimizer adapter under a public method name.

Classes

EvaluationResult(trial_id, sim_id, ...[, ...])	Objective result produced after evaluating one suggestion.
Optimizer(*args, **kwargs)	Ask/tell Protocol implemented by optimizer adapters.
ParamSuggestion(trial_id, values[, source])	Candidate parameter point proposed by an optimizer.