摘要：Context. The stellar halo holds some of the best preserved fossils of Galactic formation history that can be detected as overdensities. The detection and analysis of merger by-products within the halo enables the reconstruction of the accretion history of the Milky Way. Upcoming large-scale all-sky surveys such as Gaia and The Large Synoptic Survey Telescope (LSST) will provide a huge and rich data set, which at the same time poses challenges for automated halo debris detection.
Aims. We investigate the overdensity detection algorithm Ordering Points To Identify the Clustering Structure (OPTICS) as a method to identify tidal debris in the Galactic halo with large-scale surveys, as well as the variant FOPTICS which is capable of handling data sets with multi-dimensional uncertainty ellipsoids.
Methods. We applied OPTICS to the a simulated Galactic stellar Halo to assess the detection performance. Additionally, we tested the performance of FOPTICS is tested by introducing uncertainty ellipsoids to the 6D phase space of two test cases. We present the Jaccard index as an alternative way to test the stability of halo debris overdensity detections without the need for a local background density estimate.
Results. We optimized the OPTICS overdensity detection algorithm so that it has a slightly superlinear run-time complexity, making the method suitable for large-scale surveys. Our test on a mock galactic halo in 6D phase space shows an excellent capability to not only detect the compact dense clusters, but also the larger streams that cover a significant part of the sky. The output of OPTICS, the so-called 2D reachability diagram, proved to be a very useful tool to grasp the size, density, and substructure of the overdensities without needing to resort to complex projections of the 6D phase space. Using FOPTICS, we show the effects of introducing uncertainty ellipsoids in the 6D phase space on the retrieved tidal streams, and how the detectability of a cluster depends on whether its size and density is sufficiently large to overcome the effects of the uncertainties on the attributes.
