Exploring the parameter space of hierarchical triple black hole systems

TL;DR

This study systematically explores the parameter space of hierarchical triple black hole systems to identify conditions that facilitate mergers within a Hubble time, addressing the initial separation problem in gravitational wave astrophysics.

Contribution

It introduces a comprehensive, large-scale simulation framework with a neural network predictor to efficiently analyze triple BH mergers, incorporating relativistic effects and complex dynamics.

Findings

Merger likelihood increases with asymmetric inner binary masses.

Effective ZLK oscillations occur at large inner separations without relativistic suppression.

Outer orbit eccentricity and proximity significantly influence merger probability.

Abstract

We present a comprehensive exploration of hierarchical triple black hole (BH) systems to address the "initial separation" problem in gravitational wave astrophysics. This problem arises because isolated BH binaries must have extremely small initial separations to merge within a Hubble time via gravitational wave (GW) emission alone, separations at which their stellar progenitors would have merged prematurely. Using a modified JADE secular code incorporating GW energy loss, we systematically investigate a seven-dimensional parameter space: masses of three BHs (5-100 $M_\odot$ inner binary, 1-200 $M_\odot$ tertiary), inner/outer semimajor axes (1-200 AU and 100-10,000 AU), outer orbit eccentricity (0-0.9), and mutual inclination (40{\deg}-80{\deg}). We employed an adaptive MCMC approach sampling the merger/nonmerger transition boundary across nearly 15 million simulations. Results reveal merger-conducive regions correspond to asymmetric inner binary masses, large inner separations where von Zeipel-Lidov-Kozai (ZLK) mechanism operates effectively without relativistic precession suppression, small outer separations providing stronger perturbations, and large outer eccentricities bringing the tertiary closer at pericenter. Merger probability correlates positively with mutual inclination. We developed a classification scheme for nonmerging systems based on GW emission and ZLK oscillations. A trained neural network predicts merger outcomes with 99% ROC score and 95% accuracy (99.7% for high-confidence predictions), enabling rapid population synthesis. Validation against N-body integrations showed 87% qualitative agreement, confirming our methodology captures essential triple BH dynamics while enabling unprecedented-scale exploration of configurations resolving the initial separation problem.