accinv.loco module

class accinv.loco.Loco(model_and_jacobian_method: Union[tuple[accinv.model.Model, Type[accinv.jacobian.NumericalJacobianMethod]], tuple[accinv.model.OpticsModel, Type[accinv.jacobian.AnalyticalJacobianMethod]], tuple[accinv.model.Model, accinv.jacobian.ComputeJacobianFunction]], *, hbpms: Collection[str], hsteerers: Collection[str], vbpms: Collection[str], vsteerers: Collection[str], orm_measurement: OrmMeasurement, quadrupoles: Optional[Collection[str]] = None, multipoles: Optional[Collection[str]] = None, fit_gain_errors: bool = True, parameters: Optional[Collection[ParameterGroup]] = None, remove_coupling_blocks_from_orm: bool = False, optimizer: Optional[Optimizer] = None, compute_orm_kwargs: Optional[dict[str, Any]] = None, customize_residuals: Optional[CustomizeResiduals] = None, customize_jacobian: Optional[CustomizeJacobian] = None)

Bases: object

Main class for LOCO (linear optics from closed orbits) inverse modeling.

Note

The order in which fitting parameters are appended from shorthand arguments is the same in which they appear in the function signature, i.e. quadrupoles followed by multipoles and gain errors. This is the same order in which the resulting parameter estimates are reported.

Note

Any element or parameter names should be given in lowercase. This is the convention used by the Model backends. While lowercase conversion is ensured for the shorthand arguments (e.g. quadrupoles) this is not done for the parameters specified via parameters.

Parameters

  • model_and_jacobian_method – The accelerator model and the Jacobian method to be used for computations.

  • hbpms – Names of horizontal BPMs.

  • hsteerers – Names of horizontal steerers.

  • vbpms – Names of vertical BPMs.

  • vsteerers – Names of vertical steerers.

  • orm_measurement – The measured ORM and uncertainty to be fitted against the model.

  • quadrupoles – Names of thick quadrupoles which are added as parameters to the fitting. This is a shorthand argument. Internally, the given names are converted to Quadrupoles.

  • multipoles – Names of thin multipoles which are then added as parameters to the fitting. This is a shorthand argument. Internally, the given names are converted to Multipoles.

  • fit_gain_errors – Flag to indicate whether BPM and steerer gain errors should be fitted. If true (the default), then those gain error parameters are appended to the specified parameters which are not present there yet.

  • parameters – The parameters to be fitted. If fit_gain_errors is true, then the corresponding parameters will be added automatically, i.e. they should not appear here.

  • remove_coupling_blocks_from_orm – (optional, default: False) If true, only the main diagonal blocks of the ORM will be used for fitting. This realized by modifying the residuals and Jacobian similar to customize_residuals and customize_jacobian. The selection of main diagonal blocks is inserted before these custom functions which will only receive the main diagonal blocks as arguments.

  • optimizer – The choice of optimizer for the fitting procedure (defaults to LeastSquares).

  • compute_orm_kwargs – Additional keyword arguments for the ORM computation via Model.compute_orm(). This includes the kicks keyword argument which will default to Loco.DEFAULT_STEERER_KICKS if not specified.

  • customize_residuals – Callable which receives the current parameter guess and the corresponding residuals to return a customized version of the residuals.

  • customize_jacobian – Callable which receives the current parameter guess and the corresponding Jacobian to return a customized version of the Jacobian.

DEFAULT_STEERER_KICKS = [0.0001]
build_wrapper() → Wrapper

Build the Wrapper instance for this LOCO setup.

check_missing_elements(elements: Mapping[str, Collection[str]]) → None

Check if any of the specified elements are missing in the model.

Parameters

elements – Maps arbitrary labels to names of elements.

Raises

MissingElementsError – If any of the specified elements are missing.

run(*, guess: ~typing.Optional[<MagicMock name='mock.ndarray' id='139857761889296'>] = None, maxiter: ~typing.Optional[int] = None) → Result

Start the inverse modeling and return the result.

Parameters

  • guess – Optional initial guess for the parameter values (defaults to zero).

  • maxiter – Maximum number of iterations in terms of the Jacobian evaluation.

Returns

Result object which contains various information about the fitting result; result.x are the final parameter values.

exception accinv.loco.MissingElementsError(type_: str, names: Collection[str], /)

Bases: Exception

class accinv.loco.OrmMeasurement(orm: <MagicMock name='mock.NDArray.__getitem__()' id='139857758038624'>, uncertainty: <MagicMock name='mock.NDArray.__getitem__()' id='139857758038624'>)

Bases: object

This class represents the data for an ORM measurement.

orm: <MagicMock name='mock.NDArray.__getitem__()' id='139857758038624'>
uncertainty: <MagicMock name='mock.NDArray.__getitem__()' id='139857758038624'>