saealib.GlobalSurrogateManager#

class saealib.GlobalSurrogateManager(surrogate, training_set=None, accuracy_evaluator=None)[source]#

Bases: SurrogateManager

Surrogate manager that fits once on the full archive.

Fits the surrogate on all archived solutions, then predicts and scores all candidates in a single batch. Suitable when global approximation quality is sufficient.

Parameters:

surrogate (Surrogate) – Surrogate model instance.
training_set (TrainingSet or None) – Strategy object for building training data. Defaults to ArchiveObjectiveSet().
accuracy_evaluator (AccuracyEvaluator or None) – If provided, fit() computes accuracy metrics after each fit and stores them in last_accuracy.

Methods

`__init__`
`contract`	Return the global-manager contract.
`fit`	Fit the surrogate on the full archive.
`on_generation_end`	End-of-generation hook; override to update internal state.
`predict`	Fit on full archive, then predict all candidates at once.
`with_on_generation_end`	Return a copy of this manager with `fn` appended to the hook.

Method Details

GlobalSurrogateManager.__init__(surrogate, training_set=None, accuracy_evaluator=None)[source]#

Parameters:

surrogate (Surrogate)
training_set (TrainingSet | None)
accuracy_evaluator (AccuracyEvaluator | None)

GlobalSurrogateManager.contract()[source]#

Return the global-manager contract.

Return type:: ComponentContract

GlobalSurrogateManager.fit(archive, ctx=None)[source]#

Fit the surrogate on the full archive.

If an accuracy_evaluator was supplied at construction, accuracy metrics are computed immediately after fitting and stored in last_accuracy.

Parameters:

archive (Archive)
ctx (OptimizationState | None)

Return type:

None

GlobalSurrogateManager.on_generation_end(gen, archive, ctx=None)#

End-of-generation hook; override to update internal state.

Parameters:

gen (int)
archive (Archive)
ctx (OptimizationState | None)

Return type:

None

GlobalSurrogateManager.predict(candidates_x, archive, ctx=None, *, refit=True)[source]#

Fit on full archive, then predict all candidates at once.

Parameters:

candidates_x (np.ndarray)
archive (Archive)
ctx (OptimizationState | None)
refit (bool)

Return type:

SurrogatePrediction

GlobalSurrogateManager.with_on_generation_end(fn)#

Return a copy of this manager with fn appended to the hook.

Parameters:: fn (callable) – fn(gen, archive, ctx) -> None
Returns:: Shallow copy with the hook registered.
Return type:: SurrogateManager