Catalina Real-time Survey Research Articles

Imbalance learning for variable star classification

ABSTRACT The accurate automated classification of variable stars into their respective subtypes is difficult. Machine learning–based solutions often fall foul of the imbalanced learning problem, which causes poor generalization performance in practice, especially on rare variable star subtypes. In previous work, we attempted to overcome such deficiencies via the development of a hierarchical machine learning classifier. This ‘algorithm-level’ approach to tackling imbalance yielded promising results on Catalina Real-Time Survey (CRTS) data, outperforming the binary and multiclass classification schemes previously applied in this area. In this work, we attempt to further improve hierarchical classification performance by applying ‘data-level’ approaches to directly augment the training data so that they better describe underrepresented classes. We apply and report results for three data augmentation methods in particular: Randomly Augmented Sampled Light curves from magnitude Error (RASLE), augmenting light curves with Gaussian Process modelling (GpFit) and the Synthetic Minority Oversampling Technique (SMOTE). When combining the ‘algorithm-level’ (i.e. the hierarchical scheme) together with the ‘data-level’ approach, we further improve variable star classification accuracy by 1–4 per cent. We found that a higher classification rate is obtained when using GpFit in the hierarchical model. Further improvement of the metric scores requires a better standard set of correctly identified variable stars, and perhaps enhanced features are needed.

A Magellanic origin for the Virgo sub-structure

Iorio et al. (2018) mapped out the Milky Way halo using a sample of RR Lyrae stars drawn from a cross-match of Gaia with 2MASS. We investigate the significant residual in their model which we constrain to lie at Galactocentric radii $12<R<27\;\mathrm{kpc}$ and extend over $2600\;\mathrm{deg}^2$ of the sky. A counterpart of this structure exists in both the Catalina Real Time Survey and the sample of RR Lyrae variables identified in Pan-STARRS by Hernitschek et al. (2016), demonstrating that this structure is not caused by the spatial inhomogeneity of Gaia. The structure is likely the Virgo Stellar Stream and/or Virgo Over-Density. We show the structure is aligned with the Magellanic Stream and suggest that it is either debris from a disrupted dwarf galaxy that was a member of the Vast Polar Structure or that it is SMC debris from a tidal interaction of the SMC and LMC $3\;\mathrm{Gyr}$ ago. If the latter then the sub-structure in Virgo may have a Magellanic origin.

SPECTROSCOPY AND PHOTOMETRY OF CATACLYSMIC VARIABLE CANDIDATES FROM THE CATALINA REAL TIME SURVEY

The Catalina Real Time Survey (CRTS) has found over 500 cataclysmic variable (CV) candidates, most of which were previously unknown. We report here on followup spectroscopy of 36 of the brighter objects. Nearly all the spectra are typical of CVs at minimum light. One object appears to be a flare star, while another has a spectrum consistent with a CV but lies, intriguingly, at the center of a small nebulosity. We measured orbital periods for eight of the CVs, and estimated distances for two based on the spectra of their secondary stars. In addition to the spectra, we obtained direct imaging for an overlapping sample of 37 objects, for which we give magnitudes and colors. Most of our new orbital periods are shortward of the so-called period gap from roughly 2 to 3 hours. By considering the cross-identifications between the Catalina objects and other catalogs such as the Sloan Digital Sky Survey, we argue that a large number of cataclysmic variables remain uncatalogued. By comparing the CRTS sample to lists of previously-known CVs that CRTS does not recover, we find that the CRTS is biased toward large outburst amplitudes (and hence shorter orbital periods). We speculate that this is a consequence of the survey cadence.

Characterization of Dwarf Novae Using SDSS Colors

Abstract We have developed a method for estimating the orbital periods of dwarf novae from the Sloan Digital Sky Survey (SDSS) colors in quiescence using an artificial neural network. For typical objects below the period gap with sufficient photometric accuracy, we were able to estimate the orbital periods with accuracy to a 1 $\sigma$ error of 22%. The error of the estimation is worse for systems with longer orbital periods. We have also developed a neural-network-based method for categorical classification. This method has proven to be efficient in classifying objects into three categories (WZ Sge type, SU UMa type, and SS Cyg/Z Cam type), and works for very faint objects to a limit of g$=$ 21 mag. Using this method, we have investigated the distribution of the orbital periods of dwarf novae from a modern transient survey (Catalina Real-Time Survey). Using a Bayesian analysis developed by Uemura et al. (2010, PASJ, 62, 613), we have found that the present sample tends to give a flatter distribution to the shortest period and a shorter estimate of the period minimum, which may have resulted from uncertainties in the neural-network analysis and photometric errors. We also provide estimated orbital periods, estimated classifications, and supplemental information on known dwarf novae with the quiescent SDSS photometry.