A Machine Learns to Predict the Stability of Tightly Packed Planetary Systems

TL;DR

This paper demonstrates that machine learning, specifically XGBoost, can accurately and efficiently predict the dynamical stability of tightly packed planetary systems, significantly reducing computational costs compared to traditional N-body simulations.

Contribution

It introduces a machine learning approach to classify planetary system stability, offering a faster alternative to N-body simulations and enabling broader analysis of exoplanet data.

Findings

XGBoost achieves high accuracy in stability prediction.

Machine learning is 1000 times faster than N-body simulations.

Potential application to large exoplanet datasets like TESS.

Abstract

The requirement that planetary systems be dynamically stable is often used to vet new discoveries or set limits on unconstrained masses or orbital elements. This is typically carried out via computationally expensive N-body simulations. We show that characterizing the complicated and multi-dimensional stability boundary of tightly packed systems is amenable to machine learning methods. We find that training an XGBoost machine learning algorithm on physically motivated features yields an accurate classifier of stability in packed systems. On the stability timescale investigated ($10^7$ orbits), it is 3 orders of magnitude faster than direct N-body simulations. Optimized machine learning classifiers for dynamical stability may thus prove useful across the discipline, e.g., to characterize the exoplanet sample discovered by the upcoming Transiting Exoplanet Survey Satellite (TESS). This proof of concept motivates investing computational resources to train algorithms capable of predicting stability over longer timescales and over broader regions of phase space.