Bases: object
Ordered collection of calibrated parameters.
The space preserves declaration order, exposes transformed bounds for
optimizers, and can describe one candidate as serializable metadata for
persistence and reporting.
- Parameters:
parameters (Iterable[CalibParameter])
-
describe_values(values)[source]
Return serializable parameter metadata for one candidate.
- Parameters:
values (Mapping[str, float])
- Return type:
dict[str, dict[str, Any]]
-
default_prior_mean_std()[source]
Return Normal-prior vectors in transformed space when declared.
- Return type:
tuple[list[float], list[float]] | None
-
classmethod from_toml_mapping(declarations, *, annotations=None)[source]
Build a space from [calibration.parameters] TOML section.
annotations may provide defaults harvested from Pydantic
Calibrable hints keyed by parameter name. TOML overrides win.
- Parameters:
-
- Return type:
ParameterSpace
