Random Forests as a viable method to select and discover high redshift quasars

Abstract

We present a method of selecting quasars up to redshift $\approx$ 6 with random forests, a supervised machine learning method, applied to Pan-STARRS1 and WISE data. We find that, thanks to the increasing set of known quasars we can assemble a training set that enables supervised machine learning algorithms to become a competitive alternative to other methods up to this redshift. We present a candidate set for the redshift range 4.8 to 6.3 which includes the region around z = 5.5 where quasars are difficult to select due to photometric similarity to red and brown dwarfs. We demonstrate that under our survey restrictions we can reach a high completeness ($66\pm 7\%$ below redshift 5.6 / $83_{-9}^{+6}\%$ above redshift 5.6) while maintaining a high selection efficiency ($78_{-8}^{+10}\%$ / $94_{-8}^{+5}\%$). Our selection efficiency is estimated via a novel method based on the different distributions of quasars and contaminants on the sky. The final catalog of 515 candidates includes 225 known quasars. We predict the candidate catalog to contain an additional $148_{-33}^{+41}$ new quasars below redshift 5.6 and $45_{-8}^{+5}$ above and make the catalog publicly available. Spectroscopic follow-up observations of 37 candidates lead us to discover 20 new high redshift quasars (18 at $4.6\le z\le 5.5$, 2 $z\sim 5.7$). These observations are consistent with our predictions on efficiency. We argue that random forests can lead to higher completeness because our candidate set contains a number of objects that would be rejected by common color cuts, including one of the newly discovered redshift 5.7 quasars.

Accepted by AJ

References in corpus (16)

• Array Programming with NumPy
• The NumPy array: a structure for efficient numerical computation
• astroquery: An Astronomical Web-Querying Package in Python
• Observational constraints on Cosmic Reionization
• The Sloan Digital Sky Survey Quasar Catalog: Sixteenth Data Release
• Model-independent evidence in favor of an end to reionization by z~6
• The Pan-STARRS1 distant z>5.6 quasar survey: more than 100 quasars within the first Gyr of the universe
• The Final SDSS High-Redshift Quasar Sample of 52 Quasars at z>5.7
• Think Outside the Color Box: Probabilistic Target Selection and the SDSS-XDQSO Quasar Targeting Catalog
• A Survey of z~6 Quasars in the SDSS Deep Stripe: I. a Flux-Limited Sample at z_{AB}<21
• Constraining the radio-loud fraction of quasars at z>5.5
• The Subaru high-z quasar survey: discovery of faint z~6 quasars
• Quasar and galaxy classification in Gaia Data Release 2
• Efficient Selection of Quasar Candidates Based on Optical and Infrared Photometric Data Using Machine Learning
• The Extremely Luminous Quasar Survey (ELQS) in the SDSS footprint I.: Infrared Based Candidate Selection
• The Extremely Luminous Quasar Survey in the Pan-STARRS 1 Footprint (PS-ELQS)

Cited by in corpus (3)

• Discovery of 24 radio-bright quasars at $4.9\le z\le 6.6$ using low-frequency radio observations
• ULISSE: A Tool for One-shot Sky Exploration and its Application to Active Galactic Nuclei Detection
• The probabilistic random forest applied to the QUBRICS survey: improving the selection of high-redshift quasars with synthetic data