Abstract Details

Name: Vikrant Jadhav
Affiliation: Indian institute of astrophysics
Conference ID: ASI2020_354
Title : UVIT catalog of open clusters with machine learning based membership probability
Authors and Co-Authors : Vikrant Jadhav, Clara Pennock, Deepthi Prabhu, Annapurni Subramaniam, Ram Sagar
Abstract Type : Oral
Abstract Category : Stars, ISM and Galaxy
Abstract : Study of open clusters in ultraviolet provides an important means to investigate the formation and evolution of exotic hot population such as blue stragglers, sub-dwarfs and extremely low mass white dwarfs. We require membership information to study individual stars and the cluster in detail. Traditionally deconvolved Gaussian distributions in proper motion space are used to segregate cluster members from field stars. Large surveys, like Gaia DR2, provide a multitude of astrometric information which cannot be modelled by simple mathematics. Machine learning techniques can work with complex relations. AIM: To create a homogeneous catalogue of six open clusters (Berkeley67, King2, NGC2420, NGC2477, NGC2682 and NGC6940) using Ultra Violet Imaging Telescope (UVIT) imaging. METHOD: For consistent over all clusters, we used nine astrometric and photometric parameters from Gaia DR2 to determine the membership. We created a training set using Gaussian mixture modelling to segregate the cluster members in proper motion plane and then trained the probabilistic random forest (PRF), which handles noisy data, to get the membership probability. We investigated independent and derived Gaia DR2 parameters as features and their effect on the membership determination in PRF. We used the optimised feature selection to create a Gaia catalogue of cluster members. We observed six open clusters with UVIT aboard ASTROSAT in one or more filters. We created the images using CCDLAB and performed PSF photometry on all images using IRAF. We crossmatched the Gaia member catalogue with UVIT photometry to create an UV-optical catalogue with individual membership probability. RESULTS: We compare the results from supervised and unsupervised membership determining techniques. We calculate cluster parameters such as mean proper motion, mean parallax, velocity dispersion etc. Ultimately, we create an UV-optical catalogue using UVIT and Gaia DR2 photometry. We present the UV CMDs and brief analysis of UV characteristics of the open clusters.