MEASURING GRAVITATIONAL-WAVES FROM BINARY BLACK-HOLE COALESCENCES - II - THE WAVES INFORMATION AND ITS EXTRACTION, WITH AND WITHOUT TEMPLATES

We discuss the extraction of information from detected binary black ho le (BBH) coalescence gravitational waves by the ground-based interfero meters LIGO and VIRGO, and by the space-based interferometer LISA. We focus on the merger phase that occurs after the gradual inspiral and b efore the ringdown. Our results are (i) if numerical relativity simula tions have not produced template merger waveforms before BBH events ar e detected, one can study the merger waves using simple band-pass filt ers. For BBHs smaller than about 40M(.) detected via their inspiral wa ves, the band-pass filtering signal-to-noise ratio indicates that the merger waves should typically be just barely visible in the noise for initial and advanced LIGO interferometers. (ii) We derive an optimized maximum-likelihood method for extracting a best-fit merger waveform f rom the noisy detector output; one ''perpendicularly projects'' this o utput onto a function space (specified using wavelets) that incorporat es our (possibly sketchy) prior knowledge of the waveforms. An extensi on of the method allows one to extract the BBH's two independent wavef orms from outputs of several interferometers. (iii) We propose a compu tational strategy for numerical relativists to pursue, if they success fully produce computer codes for generating merger waveforms, but if r unning the codes is too expensive to permit an extensive survey of the merger parameter space. In this case, for LIGO-VIRGO data analysis pu rposes, it would be advantageous to do a coarse survey of the paramete r space aimed at exploring several qualitative issues and at determini ng the ranges of the several key parameters which we describe. (iv) A complete set of templates could be used to test the nonlinear dynamics of general relativity and to measure some of the binary's parameters via matched filtering. We estimate the number of bits of information o btainable from the merger waves (about 10-60 for LIGO-VIRGO, up to 200 for LISA), estimate the information loss due to template numerical er rors or sparseness in the template grid, and infer approximate require ments on template accuracy and spacing.