The eROSITA Final Equatorial-Depth Survey (eFEDS). The stellar counterparts of eROSITA sources identified by machine learning and Bayesian algorithms

Abstract

Stars are ubiquitous X-ray emitters and will be a substantial fraction of the X-ray sources detected in the on-going all-sky survey performed by the eROSITA instrument aboard the Spectrum Roentgen Gamma (SRG) observatory. We use the X-ray sources in the eROSITA Final Equatorial-Depth Survey (eFEDS) field observed during the SRG performance verification phase to investigate different strategies to identify the stars among other source categories. We focus here on Support Vector Machine (SVM) and Bayesian approaches, and our approaches are based on a cross-match with the Gaia catalog, which will eventually contain counterparts to virtually all stellar eROSITA sources. We estimate that 2060 stars are among the eFEDS sources based on the geometric match distance distribution, and we identify the 2060 most likely stellar sources with the SVM and Bayesian methods, the latter being named HamStars in the eROSITA context. Both methods reach completeness and reliability percentages of almost 90%, and the agreement between both methods is, incidentally, also about 90%. Knowing the true number of stellar sources allowed us to derive association probabilities $p_{ij}$ for the SVM method similar to the Bayesian method so that one can construct samples with defined completeness and reliability properties using appropriate cuts in $p_{ij}$. The thus identified stellar sources show the typical characteristics known for magnetically active stars, specifically, they are generally compatible with the saturation level, show a large spread in activity for stars of spectral F to G, and have comparatively high fractional X-ray luminosities for later spectral types.