Seminario DAS: Machine Learning Estimate of the Interstellar Physical Conditions

Speaker: Dr. Miguel Carvajal
Affiliation: Professor, CSIC-UHU, Universidad de Huelva

Abstract: We have explored the effect of the interstellar physical conditions on the profiles of the molecular (sub)mm wave lines. This analysis has led us to use techniques of Machine Learning (ML) to estimate the physical conditions of the interstellar medium (ISM). Hence, we have simulated a comprehensive set of molecular spectral data at different ISM physical conditions, collecting the information of their profiles in the input variables used in ML approaches. ML algorithms have been trained and tested using these simulations and their results are compared with the observed spectral lines to estimate the physical conditions of the ISM source. In particular, we have considered, as an illustrative example, the spectral lines of the two isomers HCN and HNC because they were detected in many interstellar sources and have been proven suitable to determine the gas temperature and the evolution of an interstellar object. The joint analysis of the spectral lines of the two isomers HNC and HCN, using ML approaches, has permitted us to determine the relative abundance HNC/HCN and the excitation temperature of the APEX detections of these isomers towards the source RCrA IRS7B. This work is a new application of ML approaches in the field of molecular astronomy and astrochemistry.