Quadruple-star systems are not always nested triples: a machine learning approach to dynamical stability

Pavan Vynatheya, Rosemary A. Mardling, Adrian S. Hamers

Research output: Contribution to journalArticleResearchpeer-review

1 Citation (Scopus)


The dynamical stability of quadruple-star systems has traditionally been treated as a problem involving two 'nested' triples which constitute a quadruple. In this novel study, we employed a machine learning algorithm, the multilayer perceptron (MLP), to directly classify 2 + 2 and 3 + 1 quadruples based on their stability (or long-term boundedness). The training data sets for the classification, comprised of 5 × 105 quadruples each, were integrated using the highly accurate direct N-body code mstar. We also carried out a limited parameter space study of zero-inclination systems to directly compare quadruples to triples. We found that both our quadruple MLP models perform better than a 'nested' triple MLP approach, which is especially significant for 3 + 1 quadruples. The classification accuracies for the 2 + 2 MLP and 3 + 1 MLP models are 94 and 93 per cent, respectively, while the scores for the 'nested' triple approach are 88 and 66 per cent, respectively. This is a crucial implication for quadruple population synthesis studies. Our MLP models, which are very simple and almost instantaneous to implement, are available on Github, along with python3 scripts to access them.

Original languageEnglish
Pages (from-to)2388-2398
Number of pages11
JournalMonthly Notices of the Royal Astronomical Society
Issue number2
Publication statusPublished - Oct 2023


  • binaries: general
  • gravitation
  • stars: kinematics and dynamics

Cite this