Autoregressive Planet Search: Application to the Kepler Mission

Abstract

The 4 yr light curves of 156,717 stars observed with NASA’s Kepler mission are analyzed using the autoregressive planet search (ARPS) methodology described by Caceres et al. The three stages of processing are maximum-likelihood ARIMA modeling of the light curves to reduce stellar brightness variations, constructing the transit comb filter periodogram to identify transit-like periodic dips in the ARIMA residuals, and Random Forest classification trained on Kepler team confirmed planets using several dozen features from the analysis. Orbital periods between 0.2 and 100 days are examined. The result is a recovery of 76% of confirmed planets, 97% when period and transit depth constraints are added. The classifier is then applied to the full Kepler data set; 1004 previously noticed and 97 new stars have light-curve criteria consistent with the confirmed planets, after subjective vetting removes clear false alarms and false positive cases. The 97 Kepler ARPS candidate transits mostly have periods of P < 10 days; many are ultrashort period hot planets with radii <1% of the host star. Extensive tabular and graphical output from the ARPS time series analysis is provided to assist in other research relating to the Kepler sample.

Exoplanet validation with machine learning: 50 new validated Kepler planets

“…With the advent of large datasets such work is increasingly common, and classifiers have been built for Kepler (Armstrong et al 2018;Chaushev et al 2019), and WASP (Schanche et al 2018). For the Kepler dataset, Caceres et al (2019) built a random forest model to find good candidates among the results from their 'autoregressive planet search' algorithm, achieving an AUC of 0.997 in classifying planet candidates against false positives. Shallue & Vanderburg (2018) used a convolutional neural net for a similar purpose, achieving an AUC of 0.988 and again aiming to separate candidates from false positives using a different planet search method.…”

“…Past efforts to classify candidates in transit surveys with machine learning have been made, using primarily random forests (McCauliff et al 2015;Armstrong et al 2018;Schanche et al 2018;Caceres et al 2019) and convolutional neural nets (Shallue & Vanderburg 2018;Ansdell et al 2018;Dattilo et al 2019;Chaushev et al 2019;Yu et al 2019;Osborn et al 2019). To date these have all focused on identifying FPs or ranking candidates within a survey.…”

Non-periodic type I Be/X-ray binary outbursts

“…As a result, we do not try to explore all of the parameter space to find the best fitting model to the observed light curve. Further we do not extract the exact periodicities from the data but leave that to future work (see for example Feigelson et al 2018;Caceres et al 2019). Instead, in this Section we simply consider a test case of a hydrodynamical Be star disc simulation to show the principle that non-periodic outbursts are possible in a misaligned viscous disc model.…”

Searching the Entirety of Kepler Data. I. 17 New Planet Candidates Including One Habitable Zone World

“…So far, independent Kepler searches have focused on only a subset of light-curves (e.g., Kunimoto et al 2018;Shallue & Vanderburg 2018) or a more limited range of orbital periods than examined by the Kepler team (e.g., Wang et al 2015;Caceres et al 2019). In this paper, we present an independent systematic search of the entirety of Kepler data (∼ 200, 000 stars) for planets, using the same three-transit minimum detection criteria as the Kepler team.…”