Dormant-Black-Holes-in-Binary-Systems

With the advent of gravitational wave astrophysics and advances in precise astrometry of numerous stellar sources, the search for compact objects is thriving. Rarely seen massive binaries containing a compact object play a vital role in the evolution towards compact object mergers. With the Gaia Data Release 3 (DR3), the first Gaia astrometric orbital solutions for binary sources have become available, revealing a large amount of such binary candidates. In the present notebook, we are interested in the rare cases in which a black hole (BH) is evolving in a binary system with a Main Sequence star (MS). Since the systems are extremely difficult to observe, we deal with Binary System Evolution (BSE) simulations, obtained via the SEVN (Stellar EVolution for N-body), a rapid binary population-synthesis code. In this way we can compare the few candidates found with a large synthetic dataset. It gets as input the initial conditions of stars or binaries (masses, spin, semi-major axis, eccentricity, etc.) and evolves them. Stellar evolution is calculated by interpolating pre-computed sets of PARSEC stellar tracks (Bressan et al., 2012). On the other hand, binary evolution is implemented by means of analytic and semi-analytic prescriptions. The implementation proves to be flexible so that every model can be easily changed or updated.We aim to understand what kind of processes these systems are likely to experience during their lifetime: using machine learning techniques, such as Deep Neural Network and XGBoost, we tried to infer and label the properties of the known sources and understand the importance that each feature has had in the evolution of systems. We tried, moreover, to match some of the candidates to the simulated systems in order to retrieve the guessed full evolution history of those candidates.