Revisit on "Ruling out chaos in compact binary systems"

TL;DR

This paper investigates chaos in relativistic compact binary systems using invariant Lyapunov exponents and fast Lyapunov indicators, clarifying previous conflicting results and confirming the presence of chaos in realistic scenarios.

Contribution

It demonstrates the importance of invariant chaos indicators in relativistic systems and clarifies discrepancies in earlier studies by analyzing treatment differences of Lyapunov exponents.

Findings

Chaos is confirmed in 2PN approximation of compact binaries.

Invariant Lyapunov exponents depend on treatment methods, not rescaling.

Fast Lyapunov indicator effectively detects chaos in relativistic binaries.

Abstract

Full general relativity requires that chaos indicators should be invariant in various spacetime coordinate systems for a given relativistic dynamical problem. On the basis of this point, we calculate the invariant Lyapunov exponents (LEs) for one of the spinning compact binaries in the conservative second post-Newtonian (2PN) Lagrangian formulation without the dissipative effects of gravitational radiation, using the two-nearby-orbits method with projection operations and with coordinate time as an independent variable. It is found that the actual source leading to zero LEs in one paper [J. D. Schnittman and F. A. Rasio, Phys. Rev. Lett. 87, 121101 (2001)] but to positive LEs in the other [N. J. Cornish and J. Levin, Phys. Rev. Lett. 89, 179001 (2002)] does not mainly depend on rescaling, but is due to two slightly different treatments of the LEs. It takes much more CPU time to obtain the stabilizing limit values as reliable values of LEs for the former than to get the slopes (equal to LEs) of the fit lines for the latter. Due to coalescence of some of the black holes, the LEs from the former are not an adaptive indicator of chaos for comparable mass compact binaries. In this case, the invariant fast Lyapunov indicator (FLI) of two-nearby orbits, as a very sensitive tool to distinguish chaos from order, is worth recommending. As a result, we do again find chaos in the 2PN approximation through different ratios of FLIs varying with time. Chaos cannot indeed be ruled out in real binaries.