IoW_20200514 - Gaia
Image of the Week |
Machine-learning techniques reveal hundreds of open clusters in Gaia data
|
|
|
|
Distribution of the open clusters in Galacto-centric coordinates R-Z (left) and in Helio-centric coordinates X-Y (right). Previously known open clusters are shown with red dots (left) or with the density map in red (right) (Cantat-Gaudin et al. 2018, 2019a and Castro-Ginard et al 2018, 2019). The black dots represent the newly found open clusters (Castro-Ginard et al. 2020). The sizes of the dots are proportional to the number of member of each cluster. Image credit: Castro-Ginard et al. 2020 |
|
Open clusters are groups of gravitationally bound stars that were formed in the same event – so they have the same chemical composition and age – and share a common position and proper motion. Those open clusters are fundamental objects in galaxies, and key for the understanding of the structure and evolution of the Milky Way. While young open clusters allow researchers to trace the star forming regions and to understand the star forming mechanisms, intermediate and old open clusters inform about the stellar processes and evolution of the Galactic disc. The study and search for open clusters has been boosted by the second release of the Gaia mission data (Gaia DR2). Since its publication, several studies have been finding new open clusters, but they were computationally limited to analyzing particular regions of the galactic disc, or dividing the search areas into smaller ones with a limited number of stars. A team of researchers, led by Alfred Castro-Ginard, has been developing a new methodology based on Big Data approaches to analyse the whole Gaia DR2 in the search for these objects. Castro-Ginard explains, “Before Gaia, we didn’t have a homogenous methodology to study and detect those objects, because we didn’t have such a big and precise data catalogue. That’s why we chose a machine-learning based method, which automatizes and allows the study of a big volume of data.” They used the machine-learning based methodology to search for overdensities across the whole Galactic disc, using an unsupervised clustering algorithm, which pointed to several overdensities as plausible candidates for open clusters. Then, they confirmed those candidates as open clusters through a deep learning artificial neural network, which recognized isochrone patterns in the colour and magnitude. “Before this methodology, there were around 1200 open clusters confirmed by Gaia”, says Castro-Ginard. ”Using Gaia’s data and our methodology, we have found more than 650 new clusters. This has improved and increased the catalogue, which now contains more than 2000 open clusters.” Stefan Jordan and Toni Sagristà represented the whole Gaia DR2 open cluster catalogue in a Gaia Sky visualisation. The video shows a flyby through the Galaxy, highlighting the open clusters known before and after Gaia DR2 and the differences between both catalogues. The flyby stops at UBC 274, showing that positions and proper motions of its stars are compatible with each other, and also showing that this cluster is being disrupted due to tidal forces of the Milky Way.
Video credit: ESA/Gaia/DPAC/Stefan Jordan & Toni Sagristà Sellés with Gaia Sky. Narrator voice produced with the free text-to-speech converter https://ttsmp3.com. Music credit: Concentration Kevin MacLeod (incompetech.com). Licensed under Creative Commons: By Attribution 3.0 License.
Background reading:
|
|
Credits: ESA/Gaia/DPAC, Alfred Castro-Ginard, Eduard Masana, Xavier Luri from the University of Barcelona, and Stefan Jordan and Toni Sagristà Sellés from the University of Heidelberg. [Published: 14/05/2020] |
Image of the Week Archive