pyfebiopt.optimize.optimizers

Adapters that bridge scipy optimizers to the engine interface.

Attributes

BoundsLike
Callback
ResidualFunction
ScalarObjective

Classes

OptimizerAdapter	Abstract interface implemented by optimizer adapters.
ScipyLeastSquaresAdapter	Adapter that wraps scipy.optimize.least_squares().
ScipyMinimizeAdapter	Adapter that wraps scipy.optimize.minimize().
ParallelObjectiveMap	Thread-backed map callable compatible with SciPy differential_evolution.
ScipyDifferentialEvolutionAdapter	Adapter that wraps scipy.optimize.differential_evolution().

Module Contents

pyfebiopt.optimize.optimizers.BoundsLike

pyfebiopt.optimize.optimizers.Callback

pyfebiopt.optimize.optimizers.ResidualFunction

pyfebiopt.optimize.optimizers.ScalarObjective

class pyfebiopt.optimize.optimizers.OptimizerAdapter

Abstract interface implemented by optimizer adapters.

optimizer_name: str = ''

classmethod from_options(options: collections.abc.Mapping[str, Any] | None) → OptimizerAdapter

Build adapter instance from settings mapping.

Returns:

Adapter instance configured with the provided options.

abstractmethod minimize(fun: ResidualFunction, jac: collections.abc.Callable[[numpy.ndarray], numpy.ndarray] | None, phi0: numpy.ndarray, bounds: BoundsLike, callbacks: collections.abc.Iterable[Callback] | None = None, candidate_fun: ResidualFunction | None = None) → tuple[numpy.ndarray, dict[str, object]]

Minimise the objective using the configured optimizer.

Returns:

Tuple of optimal vector and optimizer metadata dictionary.

static build(name: str, options: collections.abc.Mapping[str, Any] | None) → OptimizerAdapter

Construct an adapter by name.

Returns:

Concrete OptimizerAdapter ready for use with the engine.

class pyfebiopt.optimize.optimizers.ScipyLeastSquaresAdapter(**kwargs: Any)

Bases: OptimizerAdapter

Adapter that wraps scipy.optimize.least_squares().

Store keyword arguments forwarded to SciPy.

optimizer_name = 'least_squares'

kwargs: dict[str, Any]

minimize(fun: ResidualFunction, jac: collections.abc.Callable[[numpy.ndarray], numpy.ndarray] | None, phi0: numpy.ndarray, bounds: BoundsLike, callbacks: collections.abc.Iterable[Callback] | None = None, candidate_fun: ResidualFunction | None = None) → tuple[numpy.ndarray, dict[str, object]]

Run SciPy least_squares and report optimizer metadata.

Returns:

Tuple of optimal φ vector and metadata dictionary.

class pyfebiopt.optimize.optimizers.ScipyMinimizeAdapter(method: str = 'L-BFGS-B', **kwargs: Any)

Bases: OptimizerAdapter

Adapter that wraps scipy.optimize.minimize().

Store method name and keyword arguments.

optimizer_name = 'minimize'

classmethod from_options(options: collections.abc.Mapping[str, Any] | None) → ScipyMinimizeAdapter

Build minimize adapter and normalize method setting.

Returns:

Configured SciPy minimize adapter instance.

method = 'L-BFGS-B'

kwargs

minimize(fun: ResidualFunction, jac: collections.abc.Callable[[numpy.ndarray], numpy.ndarray] | None, phi0: numpy.ndarray, bounds: BoundsLike, callbacks: collections.abc.Iterable[Callback] | None = None, candidate_fun: ResidualFunction | None = None) → tuple[numpy.ndarray, dict[str, object]]

Run SciPy minimize and report optimizer metadata.

Returns:

Tuple of optimal φ vector and metadata dictionary.

class pyfebiopt.optimize.optimizers.ParallelObjectiveMap(max_workers: int)

Thread-backed map callable compatible with SciPy differential_evolution.

Create a bounded worker pool for scalar objective evaluations.

max_workers

__call__(func: collections.abc.Callable[[numpy.ndarray], float], iterable: collections.abc.Iterable[numpy.ndarray]) → list[float]

Evaluate candidates concurrently while preserving SciPy’s map order.

Returns:

Scalar objective values in the same order as the input candidates.

shutdown() → None

Release worker threads.

__enter__() → ParallelObjectiveMap

Return the map callable for context-managed use.

__exit__(*_exc_info: object) → None

Shut down worker threads when SciPy returns or raises.

class pyfebiopt.optimize.optimizers.ScipyDifferentialEvolutionAdapter(**kwargs: Any)

Bases: OptimizerAdapter

Adapter that wraps scipy.optimize.differential_evolution().

Store keyword arguments forwarded to SciPy.

optimizer_name = 'differential_evolution'

kwargs

minimize(fun: ResidualFunction, jac: collections.abc.Callable[[numpy.ndarray], numpy.ndarray] | None, phi0: numpy.ndarray, bounds: BoundsLike, callbacks: collections.abc.Iterable[Callback] | None = None, candidate_fun: ResidualFunction | None = None) → tuple[numpy.ndarray, dict[str, object]]

Run SciPy differential_evolution and report optimizer metadata.

Returns:

Tuple of optimal φ vector and metadata dictionary.