`gnn_tracking_hpo.cluster_scans`#

Module Contents#

`fixed_dbscan_scan`(...)	Convenience function for not scanning for DBSCAN hyperparameters at all.
`reduced_dbscan_scan`(...)	Convenience function for scanning DBSCAN hyperparameters with trial count

gnn_tracking_hpo.cluster_scans.fixed_dbscan_scan(graphs: numpy.ndarray, truth: numpy.ndarray, sectors: numpy.ndarray, pts: numpy.ndarray, reconstructable: numpy.ndarray, *, guide='trk.double_majority_pt1.5', epoch=None, start_params: dict[str, Any] | None = None) → gnn_tracking.postprocessing.clusterscanner.ClusterScanResult#: Convenience function for not scanning for DBSCAN hyperparameters at all.

gnn_tracking_hpo.cluster_scans.reduced_dbscan_scan(graphs: list[numpy.ndarray], truth: list[numpy.ndarray], sectors: list[numpy.ndarray], pts: list[numpy.ndarray], reconstructable: list[numpy.ndarray], *, guide='trk.double_majority_pt0.9', epoch=None, start_params: dict[str, Any] | None = None, node_mask: list[numpy.ndarray] | None = None) → gnn_tracking.postprocessing.clusterscanner.ClusterScanResult#: Convenience function for scanning DBSCAN hyperparameters with trial count that depends on the epoch (using many trials early on, then alternating between fixed and low samples in later epochs).