Extended stellar systems in the solar neighborhood IV. Meingast 1

Sebastian Ratzenboeck, Stefan Meingast, Joao Alves, Torsten Moeller, Immanuel Bomze

Context. Nearby stellar streams carry unique information on the dynamical evolution and disruption of stellar systems in the Galaxy, the mass distribution in the disk, and they provide unique targets for planet formation and evolution studies. Recently, Meingast 1, a 120 degrees stellar stream with a length of at least 400 pc, was dicovered.

Aims. We aim to revisit the Meingast 1 stream to search for new members within its currently known 400 pc extent, using Gaia DR2 data and an innovative machine learning approach.

Methods. We used a bagging classifier of one-class support vector machines with Gaia DR2 data to perform a 5D search (positions and proper motions) for new stream members. The ensemble was created by randomly sampling 2.4 million hyper-parameter realizations admitting classifiers that fulfill a set of prior assumptions. We used the variable prediction frequency resulting from the multitude of classifiers to estimate a stream membership criterion, which we used to select high-fidelity sources. We used the HR diagram and the Cartesian velocity distribution as test and validation tools.

Results. We find about 2000 stream members with high fidelity, or about an order of magnitude more than previously known, unveiling the stream's population across the entire stellar mass spectrum, from B stars to M stars, including white dwarfs. We find that, apart from being slightly more metal poor, the HRD of the stream is indistinguishable from that of the Pleiades cluster. For the mass range at which we are mostly complete, similar to 0. M-circle dot<M<similar to 4 M-circle dot, we find a normal IMF, allowing us to estimate the total mass of stream to be about 2000 M-circle dot, making this relatively young stream by far the most massive one known. In addition, we identify several white dwarfs as potential stream members.

Conclusions. The nearby Meingast 1 stream, due to its richness, age, and distance, is a new fundamental laboratory for star and planet formation and evolution studies for the poorly studied and gravitationally unbound star formation mode. We also demonstrate that one-class support vector machines can be effectively used to unveil the full stellar populations of nearby stellar systems with Gaia data.

Research Network Data Science, Department of Astrophysics, Research Group Visualization and Data Analysis, Department of Statistics and Operations Research, Research Platform Governance of digital practices
External organisation(s)
Harvard University
Astronomy & Astrophysics
No. of pages
Publication date
Peer reviewed
Austrian Fields of Science 2012
101015 Operations research, 103003 Astronomy
Portal url