Discovery of 36 eclipsing EL CVn binaries found by the Palomar Transient Factory

Abstract

We report the discovery and analysis of 36 new eclipsing EL CVn-type binaries, consisting of a core helium-composition pre-white dwarf and an early-type main-sequence companion, more than doubling the known population of these systems. We have used supervised machine learning methods to search 0.8 million lightcurves from the Palomar Transient Factory, combined with SDSS, Pan-STARRS and 2MASS colours. The new systems range in orbital periods from 0.46 to 3.8 d and in apparent brightness from ∼14 to 16 mag in the PTF R or g filters. For twelve of the systems, we obtained radial velocity curves with the Intermediate Dispersion Spectrograph at the Isaac Newton Telescope. We modelled the lightcurves, radial velocity curves and spectral energy distributions to determine the system parameters. The radii (0.3-0.7 R ) and effective temperatures (8000-17000 K) of the pre-He-WDs are consistent with stellar evolution models, but the masses (0.12-0.28 M ) show more variance than models have predicted. This study shows that using machine learning techniques on large synoptic survey data is a powerful way to discover substantial samples of binary systems in short-lived evolutionary stages.

White dwarf and subdwarf stars in the Sloan Digital Sky Survey Data Release 14

“…This leads to a large scatter, placing objects above and below the main sequence, the latter a region compatible with low metallicity main sequence stars, or interacting binary remnants (e.g. Maxted et al 2014a;Pelisoli, Kepler, & Koester 2018a;Pelisoli et al 2018bPelisoli et al , 2019van Roestel et al 2018;Wang et al 2018). Some sdAs could be very low mass white dwarfs or pre-ELMs, but with the current parallax uncertainties this cannot be confirmed.…”

Gaia Data Release 2 catalogue of extremely low-mass white dwarf candidates

“…We have not included at this point ELMs found as companions to millisecond-pulsars, because they are mostly too faint to be detectable by Gaia, or even to be fully confirmed ELMs, but a list of such objects can be found in Appendix A. We have also refrained from including systems in which (pre-)ELMs are the secondary star, the socalled EL CVn binaries (named after their prototype, e.g Maxted et al 2014;van Roestel et al 2018). These systems are dominated by the primary main sequence star, and therefore overlap with the main sequence in the Gaia HR diagram.…”

Surveying the reach and maturity of machine learning and artificial intelligence in astronomy

“…With regards to the role of ML and AI in advancing knowledge in astronomy, there was clear evidence from the sample of recent publications that discovery tasks are being performed with all of the data types: images (Ciuca & Hernández, ; Gomez Gonzalez, Absil, & Van Droogenbroeck, ; Hartley, Flamary, Jackson, Tagore, & Metcalf, ; Jacobs et al, ; Lanusse et al, ; Morello, Morris, Van Dyk, Marston, & Mauerhan, ; Pourrahmani et al, ; Wan et al, ); spectroscopy (Bu, Lei, Zhao, Bu, & Pan, ; Li et al, ); photometry (Ostrovski et al, ; Timlin et al, ; Vida & Roettenbacher, ); light curves (Armstrong et al, ; Cohen et al, ; Giles & Walkowicz, ; Hedges, Hodgkin, & Kennedy, ; Heinze et al, ; Peña et al, ; van Roestel et al, ); time‐series (Connor & van Leeuwen, ; Farah et al, ; Michilli et al, ; Morello et al, ; Pang et al, ; Tan et al, ); catalogues (Lin et al, ; Marchetti et al, ; Nguyen, Pankratius, Eckman, & Seager, ; Yan et al, ); and simulation (Kuntzer & Courbin, ; Nadler et al, ; Xu & Offner, ).…”

“…The common starting point is to apply a machine learning technique to perform a classification, regression or clustering task. Once established as being comparable to, or exceeding, a more traditional approach, machine learning can be used to forecast likely future outcomes (e.g., solar flares [Florios et al, ; Nishizuka et al, ] or coronal mass ejections from the Sun [Inceoglu et al, ]) or make new discoveries (e.g., classification schemes for stellar types permitting the identification of new candidates of rare objects as in Bu et al, , van Roestel et al, , and Zhang et al, ). The most mature disciplines move beyond classification and discovery as ends in their own to that of gaining insight—where new physical knowledge is identified, often for the first time, because a machine learning approach was used.…”