Search for OB associations in Gaia Early Data Release 3

The distribution of young stars into OB associations has long been in need of updating. High-precision Gaia Early Data Release 3 astrometry, coupled with modern machine-learning methods, allows this to be done. We have compiled a well-defined sample which includes OB stars and young open clusters (OCs), in total comprising about 47 700 objects. To break the sample down into groupings resembling associations, we applied the HDBSCAN∗ clustering algorithm. We used a Monte Carlo method to estimate the kinematic ages of the resulting clusters and the Student's t-test to assess the significance of the linear correlations between proper motions and coordinates, indicating the presence of possible cluster expansion signatures. The ages of the majority of clusters demonstrating a general expansion at a 1σ confidence level are several tens of Myr, which is in agreement with the expected ages of OB associations. We found 32 OCs which turned out to be members of the resulting groupings; their ages are consistent with one another within the uncertainties. Comparison of the clusters thus obtained with the historical composition of OB associations in the literature shows a correspondence between their positions in the Galaxy but an apparent absence of good one-to-one stellar matches. Therefore, we suggest that the historical composition of OB associations needs to be revised.