Systematic biases in parameter estimation of binary black-hole mergers

Abstract

Parameter estimation of binary black-hole merger events in gravitational-wave data relies on matched-filtering techniques which, in turn, depend on accurate model waveforms. Here we characterize the systematic biases introduced in measuring astrophysical parameters of binary black holes by applying the currently most accurate effective-one-body templates to simulated data containing nonspinning numerical-relativity waveforms. We quantify the systematic bias by using a Markov chain Monte Carlo algorithm to sample the posterior distribution function of noise-free data, and compare the offset of the maximum a priori waveform parameters (the bias) to the width of the distribution, which we refer to as the statistical error. For advanced ground-based detectors, we find that the systematic biases are well within the statistical error for realistic signal-to-noise ratios. These biases grow to be comparable to the statistical errors at high ground-based-instrument signal-to-noise ratios ($\mathrm{SNR}\ensuremath{\sim}50$), but never dominate the error budget. At the much larger signal-to-noise ratios expected for space-based detectors, these biases will become large compared to the statistical errors, but for astrophysical black hole mass estimates the absolute biases (of at most a few percent) are still fairly small.