Flux-based statistical prediction of three-body outcomes

Barak Kol*

*Corresponding author for this work

Research output: Contribution to journalArticlepeer-review

19 Scopus citations


Since Poincaré, the three-body problem is known to be chaotic and is believed to lack a general deterministic solution. Instead, decades ago a statistical solution was marked as a goal. Yet, despite considerable progress, all extant approaches display two flaws. First, probability was equated with phase space volume, thereby ignoring the fact that significant regions of phase space describe regular motion, including post-decay motion. Secondly and relatedly, an adjustable parameter, the strong interaction region, which is a sort of cutoff, was a central ingredient of the theory. This paper introduces remedies and presents for the first time a statistical prediction of decay rates, in addition to outcomes. Based on an analogy with a particle moving within a leaky container, the statistical distribution is presented in an exactly factorized form. One factor is the flux of phase-space volume, rather than the volume itself, and it is given in a cutoff-independent closed form. The other factors are the chaotic absorptivity and the regularized phase space volume. The situation is analogous to Kirchhoff’s law of thermal radiation, also known as greybody radiation. In addition, an equation system for the time evolution of the statistical distribution is introduced; it describes the decay rate statistics while accounting for sub-escape excursions. Early numerical tests indicate a leap in accuracy.

Original languageAmerican English
Article number17
JournalCelestial Mechanics and Dynamical Astronomy
Issue number4
StatePublished - Apr 2021

Bibliographical note

Publisher Copyright:
© 2021, The Author(s), under exclusive licence to Springer Nature B.V.


  • Chaos
  • Statistical mechanics
  • Three-body problem


Dive into the research topics of 'Flux-based statistical prediction of three-body outcomes'. Together they form a unique fingerprint.

Cite this