No Arabic abstract
Gaia is a European Space Agency (ESA) astrometry space mission, and a successor to the ESA Hipparcos mission. Gaias main goal is to collect high-precision astrometric data (i.e. positions, parallaxes, and proper motions) for the brightest 1 billion objects in the sky. These data, complemented with multi-band, multi-epoch photometric and spectroscopic data collected from the same observing platform, will allow astronomers to reconstruct the formation history, structure, and evolution of the Galaxy. Gaia will observe the whole sky for 5 years, providing a unique opportunity for the discovery of large numbers of transient and anomalous events, e.g. supernovae, novae and microlensing events, GRB afterglows, fallback supernovae, and other theoretical or unexpected phenomena. The Photometric Science Alerts team has been tasked with the early detection, classification and prompt release of anomalous sources in the Gaia data stream. In this paper, we discuss the challenges we face in preparing to use Gaia to search for transient phenomena at optical wavelengths.
Since July 2014, the Gaia mission has been engaged in a high-spatial-resolution, time-resolved, precise, accurate astrometric, and photometric survey of the entire sky. Aims: We present the Gaia Science Alerts project, which has been in operation since 1 June 2016. We describe the system which has been developed to enable the discovery and publication of transient photometric events as seen by Gaia. Methods: We outline the data handling, timings, and performances, and we describe the transient detection algorithms and filtering procedures needed to manage the high false alarm rate. We identify two classes of events: (1) sources which are new to Gaia and (2) Gaia sources which have undergone a significant brightening or fading. Validation of the Gaia transit astrometry and photometry was performed, followed by testing of the source environment to minimise contamination from Solar System objects, bright stars, and fainter near-neighbours. Results: We show that the Gaia Science Alerts project suffers from very low contamination, that is there are very few false-positives. We find that the external completeness for supernovae, $C_E=0.46$, is dominated by the Gaia scanning law and the requirement of detections from both fields-of-view. Where we have two or more scans the internal completeness is $C_I=0.79$ at 3 arcsec or larger from the centres of galaxies, but it drops closer in, especially within 1 arcsec. Conclusions: The per-transit photometry for Gaia transients is precise to 1 per cent at $G=13$, and 3 per cent at $G=19$. The per-transit astrometry is accurate to 55 milliarcseconds when compared to Gaia DR2. The Gaia Science Alerts project is one of the most homogeneous and productive transient surveys in operation, and it is the only survey which covers the whole sky at high spatial resolution (subarcsecond), including the Galactic plane and bulge.
Gaia Photometric Science Alerts (GPSA) publishes Gaia G magnitudes and Blue Photometer (BP) and Red Photometer (RP) low-resolution epoch spectra of transient events. 27 high-resolution spectra from Gaias Radial Velocity Spectrometer (RVS) of 12 GPSAs have also been published. These 27 RVS epoch spectra are presented next to their corresponding BP and RP epoch spectra in a single place for the first time. We also present one new RVS spectrum of a 13th GPSA that could not be published by the GPSA system. Of the 13 GPSA with RVS spectra, five are photometrically classified as unknown, five as supernovae (three as SN Ia, one as SN II, one as SN IIP), one as a cataclysmic variable, one as a binary microlensing event and one as a young stellar object. The five GPSAs classified as unknown are potential scientific opportunities, while all of them are a preview of the epoch RVS spectra that will be published in Gaias fourth data release.
Gaia is regularly producing Alerts on objects where photometric variability has been detected. The physical nature of these objects has often to be determined with the complementary observations from ground-based facilities. We have compared the list of Gaia Alerts (until 20181101) with archival LAMOST and SDSS spectroscopic data. The date of the ground-based observation rarely corresponds to the date of the Alert, but this allows at least the identification of the source if it is persistent, or the host galaxy if the object was only transient like a supernova. A list of Gaia Nuclear Transients from Kostrzewa-Rutkowska et al. (2018) has been included in this search also. We found 26 Gaia Alerts with spectra in LAMOST+SDSS labelled as stars (12 with multi-epoch spectra). A majority of them are CVs. Similarly 206 Gaia Alerts have associated spectra labelled as galaxies (49 with multi-epoch spectra). Those spectra were generally obtained on a date different from the Alert date, are mostly emission-line galaxies, leading to the suspicion that most of the Alerts were due to a SN. As for the GNT list, we found 55 associated spectra labelled as galaxies (13 with multi-epoch spectra). In two galaxies, Gaia17aal and GNTJ170213+2543, was the date of the spectroscopic observation close enough to the Alert date: we find a trace of the SN itself in their LAMOST spectrum, both classified here as a type Ia SN. The GNT sample has a higher proportion of AGNs, suggesting that some of the detected variations are also due to the AGN itself. Similar for Quasars, we found 30 Gaia Alerts but 68 GNT cases have single epoch quasar spectra, while 12 plus 23 have multi-epoch spectra. For ten out of these 35, their multi-epoch spectra show appearance or disappearance of the broad Balmer lines and also variations in the continuum, qualifying them as Changing Look Quasars.
We consider the problem of 20 questions with noise for multiple players under the minimum entropy criterion in the setting of stochastic search, with application to target localization. Each player yields a noisy response to a binary query governed by a certain error probability. First, we propose a sequential policy for constructing questions that queries each player in sequence and refines the posterior of the target location. Second, we consider a joint policy that asks all players questions in parallel at each time instant and characterize the structure of the optimal policy for constructing the sequence of questions. This generalizes the single player probabilistic bisection method for stochastic search problems. Third, we prove an equivalence between the two schemes showing that, despite the fact that the sequential scheme has access to a more refined filtration, the joint scheme performs just as well on average. Fourth, we establish convergence rates of the mean-square error (MSE) and derive error exponents. Lastly, we obtain an extension to the case of unknown error probabilities. This framework provides a mathematical model for incorporating a human in the loop for active machine learning systems.
Using Gaia DR2 data, we present an up-to-date sample of white dwarfs within 20 pc of the Sun. In total we identified 139 systems in Gaia DR2, nine of which are new detections, with the closest of these located at a distance of 13.05 pc. We estimated atmospheric parameters for all stellar remnants based on the Gaia parallaxes and photometry. The high precision and completeness of the Gaia astrometry allowed us to search for wide binary companions. We re-identified all known binaries where both components have accurate DR2 astrometry, and established the binarity of one of the nine newly identified white dwarfs. No new companions were found to previously known 20 pc white dwarfs. Finally, we estimated the local white dwarf space-density to be $(4.49pm0.38)times10^{-3}$ pc$^{-3}$, having given careful consideration to the distance-dependent Gaia completeness, which misses known objects at short distances, but is close to complete for white dwarfs near 20 pc.