classify_association_clusters#

lsst.analysis.ap.classify_association_clusters(sources1, sources2, match_radius=0.5)#

Enumerate and classify every association-disagreement cluster between two APDBs that share input diaSources.

Builds the bipartite graph whose edges are (diaSource -> its run-1 diaObject, diaSource -> its run-2 diaObject) over all diaSources the two runs have in common, runs union-find over the diaObjectIds to extract every connected component, and labels each cluster:

  • matched – one run-1 obj <-> one run-2 obj.

  • split – one run-1 obj split into multiple run-2 objs.

  • merged – multiple run-1 objs merged into one run-2 obj.

  • tangled – M run-1 objs <-> N run-2 objs, both > 1.

diaSourceIds are not stable across runs: they carry a per-catalog counter assigned in detection order, so any change to detection or measurement renumbers them. The common diaSources are therefore identified by position (nearest neighbor within match_radius, inside a single (visit, detector)) rather than by id. Sources with no counterpart in the other run are skipped.

Parameters#

sources1, sources2pandas.DataFrame

Full diaSources catalogs from runs 1 and 2 (e.g. from query.load_sources()). Each must contain diaSourceId, diaObjectId, ra, dec, visit, and detector columns.

match_radiusfloat, optional

Maximum separation in arcsec for two diaSources to be considered the same detection in both runs.

Returns#

clusterspandas.DataFrame
One row per cluster, with columns:

  • kind: matched / split / merged / tangled.

  • n_obj1, n_obj2: distinct diaObject counts per run.

  • n_sources: matched diaSource pairs in the cluster.

  • obj1_ids, obj2_ids: tuples of diaObjectIds.

  • ra, dec: mean sky position of the cluster’s diaSources (degrees).