Parameter estimation of stellar-mass black hole binaries with LISA

TL;DR

This paper evaluates LISA's capability to estimate parameters of stellar-mass black hole binaries during their inspiral phase, demonstrating high precision in certain parameters and discussing analysis methods and degeneracies.

Contribution

It introduces a Bayesian analysis framework for LISA observations of SBHBs, including an augmented Fisher-matrix approach and insights into parameter degeneracies.

Findings

Redshifted chirp mass and sky location are very well determined.

Luminosity distance typically measured within 40-60%.

Error on coalescence time reduces significantly closer to merger.

Abstract

Stellar-mass black hole binaries (SBHBs), like those currently being detected with the ground-based gravitational-wave (GW) observatories LIGO and Virgo, are also an anticipated GW source for LISA. LISA will observe them during the early inspiral stage of evolution; some of them will chirp through the LISA band and reappear some time later in the band of $3^{rd}$ generation ground-based detectors. SBHBs could serve as laboratories for testing the theory of General Relativity and inferring the astrophysical properties of the underlying population. In this study, we assess LISA's ability to infer the parameters of those systems, a crucial first step in understanding and interpreting the observation of those binaries and their use in fundamental physics and astrophysics. We simulate LISA observations for several fiducial sources and perform a full Bayesian analysis. We demonstrate and explain degeneracies in the parameters of some systems. We show that the redshifted chirp mass and the sky location are always very well determined, with typical errors below $10^{-4}$ (fractional) and $0.4 \ {\rm deg^2}$. The luminosity distance to the source is typically measured within $40-60\%$, resulting in a measurement of the chirp mass in the source frame of $\mathcal{O}(1 \%)$. The error on the time to coalescence improves from $\mathcal{O}(1 \ {\rm day})$ to $\mathcal{O}(30 \ {\rm s})$ as we observe the systems closer to their merger. We introduce an augmented Fisher-matrix analysis which gives reliable predictions for the intrinsic parameters compared to the full Bayesian analysis. Finally, we show that combining the use of the long-wavelength approximation for the LISA instrumental response together with the introduction of a degradation function at high frequencies yields reliable results for the posterior distribution when used self-consistently, but not in the analysis of real LISA data.