OutputSettings#
- class OutputSettings(save_prefit_residuals: bool = True, save_postfit_residuals: bool = True, save_covariance_history: bool = True, save_consider_covariance_history: bool = True, save_state_deviation_history: bool = True, save_kalman_gains: bool = False, save_innovation_covariances: bool = False, save_nis_statistics: bool = False, save_smoothed_solution: bool = True, solution_output_path: str | None = None, solution_output_name: str | None = 'solution', fdm_output_path: str | None = None, fdm_output_name: str | None = None, fdm_output_format: Literal['json', 'pickle'] | None = 'json', trajectory_output_format: Literal['json', 'pickle'] | None = 'json', metadata: Dict | None = None, output_dir: str | None = None, run_name: str | None = 'scenario', save_reference_trajectory: bool = True, save_estimated_trajectory: bool = True, save_trajectory_stms: bool = True, save_full_covariance: bool = True)#
Bases:
objectConfiguration for
SolutionODoutput and storage.Controls which quantities the filter stores after a run and where the JSON output files are written. All flags default to sensible values; override only what you need.
- Parameters:
- save_prefit_residualsbool
Store prefit residuals in the solution. Defaults to
True.- save_postfit_residualsbool
Store postfit residuals in the solution. Defaults to
True.- save_covariance_historybool
Store covariance evolution across measurement epochs. Defaults to
True.- save_consider_covariance_historybool
Store consider-parameter covariance evolution. Defaults to
True.- save_state_deviation_historybool
Store state deviation history across measurement epochs. Defaults to
True.- save_kalman_gainsbool
Store Kalman gain matrices at each epoch (debugging). Defaults to
False.- save_innovation_covariancesbool
Store innovation covariance matrices at each epoch (debugging). Defaults to
False.- save_nis_statisticsbool
Store Normalized Innovation Squared (NIS) statistics. Defaults to
False.- save_smoothed_solutionbool
Run the RTS/Bierman smoother after the forward pass and store the smoothed solution. Defaults to
True.- solution_output_pathstr, optional
Directory for the JSON solution file.
Nonedisables JSON output.- solution_output_namestr, optional
Base filename (without extension) for the JSON solution file. Defaults to
'solution'.- fdm_output_pathstr, optional
Directory to write per-dataset JSON filter-data files.
Nonedisables filter-data output.- fdm_output_namestr, optional
Base name prefix for filter-data files.
- fdm_output_format{“json”, “pickle”}, optional
On-disk format for the filter-data files — the per-dataset measurement records the filter rebuilds each iteration. Defaults to
"json". This does not control the trajectory sidecars; seetrajectory_output_format.- save_trajectory_stmsbool
Whether to write the STM sidecar beside each trajectory SPK. It is the largest artifact an OD run produces — one matrix per output epoch, per iteration — and the filter serves its own STM lookups from memory, so a run that does not read the sidecar back can turn it off. Defaults to
True.- trajectory_output_format{“json”, “pickle”}, optional
On-disk format for the sidecars written beside each trajectory SPK: the STM history (
*_stms.*) and the parameter / settings / mass-profile records. Defaults to"json".The STM sidecar is by far the largest artifact an OD run produces — one matrix per output epoch, growing with the square of the state size — so a finely sampled arc or an extra solve-for parameter can push it into hundreds of megabytes of text.
"pickle"writes the same content as a compact binary, which both writes and reads back considerably faster. Either format is read back transparently.- metadatadict, optional
Arbitrary user metadata (version, producer, description, etc.). Its keys are merged last into the saved
metadatablock, so they override the generated fields of the same name.- save_full_covariancebool
Write the final solution’s covariances as full matrices (
covariance_full,covariance_smooth_full,covariance_consider_full) instead of their diagonals (*_diagonals). Iteration blocks always store diagonals, and a block never holds both. Defaults toTrue.
- Attributes:
fdm_output_nameBase name prefix for filter-data files.
fdm_output_pathDirectory to write per-dataset JSON filter-data files.
metadataArbitrary user metadata (version, producer, description, etc.).
output_dirSingle directory all artifacts for this run are written under.
run_nameBase name (prefix) shared by every artifact of this run.
save_consider_covariance_historyStore consider-parameter covariance evolution.
save_covariance_historyStore covariance evolution across measurement epochs.
save_innovation_covariancesStore innovation covariance matrices at each epoch (debugging).
save_kalman_gainsStore Kalman gain matrices at each epoch (debugging).
save_nis_statisticsStore Normalized Innovation Squared (NIS) statistics.
save_postfit_residualsStore postfit residuals in the solution.
save_prefit_residualsStore prefit residuals in the solution.
save_smoothed_solutionIf
True, run the RTS/Bierman smoother after the forward pass and store the smoothed solution.save_state_deviation_historyStore state deviation history across measurement epochs.
solution_output_nameBase filename (without extension) for the JSON solution file.
solution_output_pathDirectory to save the JSON solution file.
Methods
path_for(artifact[, ext])Build a standardized
<output_dir>/<run_name>_<artifact>.<ext>path.See also
scarabaeus.SolutionODOD result container that respects these settings.
scarabaeus.FilterSettingsBundles this object with covariance and process noise.
Notes
All boolean flags default to sensible values for a standard OD run. Override only what you need; unset flags revert to their defaults when an instance is created with no arguments.
- path_for(artifact: str, ext: str = 'json') str#
Build a standardized
<output_dir>/<run_name>_<artifact>.<ext>path.- Parameters:
- artifactstr
Artifact suffix, e.g.
"solution","reference","estimated","stms","settings","parameters".- extstr
File extension without the dot. Defaults to
"json".
- Returns:
- str
The resolved path. Falls back to the current directory when
output_diris unset.
- property fdm_output_name: str | None#
Base name prefix for filter-data files. When set, files are named
{fdm_output_name}_{set_name}.json.
- property fdm_output_path: str | None#
Directory to write per-dataset JSON filter-data files.
Nonedisables filter-data output.
- property save_consider_covariance_history: bool#
Store consider-parameter covariance evolution. Defaults to
True.
- property save_covariance_history: bool#
Store covariance evolution across measurement epochs. Defaults to
True.
- property save_innovation_covariances: bool#
Store innovation covariance matrices at each epoch (debugging). Defaults to
False.
- property save_kalman_gains: bool#
Store Kalman gain matrices at each epoch (debugging). Defaults to
False.
- property save_nis_statistics: bool#
Store Normalized Innovation Squared (NIS) statistics. Defaults to
False.
- property save_smoothed_solution: bool#
If
True, run the RTS/Bierman smoother after the forward pass and store the smoothed solution. Defaults toTrue.
- property save_state_deviation_history: bool#
Store state deviation history across measurement epochs. Defaults to
True.