FilterSettings#

class FilterSettings(initial_covariance: CovarianceMatrix, initial_deviation: ndarray | None = None, process_noise: ProcessNoiseSettings | List[ProcessNoiseSettings] | None = None, measurement_sigmas: Dict[str, ArrayWUnits | float] | None = None, covariance_analysis: bool = False, output: OutputSettings | None = None, editing_method: str | None = None, editing_datasets: str | list | None = None, editing_kwargs: Dict | None = None)#

Bases: object

Comprehensive filter configuration for orbit determination.

Bundles the initial covariance, optional process noise model, measurement noise overrides, output configuration, and measurement editing settings into a single object passed to any FilterOD subclass.

Parameters:
initial_covarianceCovarianceMatrix

Initial covariance matrix \(P_0\).

initial_deviationnp.ndarray, optional

Initial state deviation \(x_0\). None means zero deviation.

process_noiseProcessNoiseSettings or list of ProcessNoiseSettings, optional

Process noise model configuration, one per noised body (see body). None disables process noise.

measurement_sigmasdict, optional

Measurement sigmas keyed by dataset name, as ArrayWUnits (converted to the dataset’s units) or bare floats (taken to be in them already). The filter sets them once, at construction, on the model that builds each dataset, so they win over the sigma the model was built with and model.sigma reports them; a sigma set on the model later wins over them. Changing this dict after construction does nothing: use FilterOD.set_measurement_sigmas(). See that method for the rules on models that build several datasets.

covariance_analysisbool

If True, run in covariance-analysis mode (no measurement updates). Defaults to False.

outputOutputSettings, optional

Output configuration. Defaults to all-default OutputSettings.

editing_methodstr, optional

Measurement editing mode: 'lasso', 'lasso_all', 'chi2', 'chi2_all', 'kurtosis', 'kurtosis_all', 'date_range', or 'interactive_date_range'. None disables editing. 'lasso', 'chi2' and 'kurtosis' open one figure per dataset in sequence; the '_all' variants open them all at once, which is the usable option on remote/browser backends.

editing_datasetsstr or list, optional

Which datasets to edit: None = all, a string = one by name, a list of strings = a specific subset.

editing_kwargsdict, optional

Extra keyword arguments forwarded to the editing method.

Attributes:
covariance_analysis

If True, run in covariance-analysis mode (no measurement updates).

editing_datasets

Which datasets to edit: None = all, a single name string, or a list of dataset name strings.

editing_kwargs

Extra keyword arguments forwarded to the editing method.

editing_method

Measurement editing mode.

initial_covariance

Initial covariance matrix \(P_0\).

initial_deviation

Initial state deviation \(x_0\).

measurement_sigmas

Measurement sigmas by dataset name, set on the models at filter construction.

output

Output configuration (OutputSettings).

process_noise

Process noise model configuration (ProcessNoiseSettings), or a list of them, one per noised body.

See also

scarabaeus.FilterOD

Base filter class that consumes this settings object.

scarabaeus.ProcessNoiseSettings

Process noise model configuration.

scarabaeus.OutputSettings

Controls what the solution stores and writes.

Notes

Pass a single FilterSettings instance to any FilterOD subclass constructor. When output is omitted, an all-default OutputSettings is created automatically.

property covariance_analysis: bool#

If True, run in covariance-analysis mode (no measurement updates).

property editing_datasets: str | list | None#

Which datasets to edit: None = all, a single name string, or a list of dataset name strings.

property editing_kwargs: Dict | None#

Extra keyword arguments forwarded to the editing method. E.g. {"sigma_scale": 4.0} for 'chi2', or {"t_start": ..., "t_end": ..., "action": "flag"} for 'date_range'.

property editing_method: str | None#

Measurement editing mode. One of 'lasso', 'chi2', 'date_range', 'interactive_date_range', or None.

property initial_covariance: CovarianceMatrix#

Initial covariance matrix \(P_0\).

property initial_deviation: ndarray | None#

Initial state deviation \(x_0\). None means zero deviation.

property measurement_sigmas: Dict[str, ArrayWUnits | float] | None#

Measurement sigmas by dataset name, set on the models at filter construction.

property output: OutputSettings#

Output configuration (OutputSettings). Always non-None after construction — defaults to an all-default OutputSettings.

property process_noise: ProcessNoiseSettings | List[ProcessNoiseSettings] | None#

Process noise model configuration (ProcessNoiseSettings), or a list of them, one per noised body.