Surrogate models of gravitational waveforms from numerical relativity simulations of black hole mergers
Jonathan Blackman (Caltech)
November 10, 2016
Abstract: GW150914 was the first detection of gravitational waves from a binary blackholemerger, bringing us into the era of gravitational wave astronomy. From suchgravitational wave detections, we can put constraints on deviations fromgeneralrelativity (GR), as well as measure the masses and spins of the black holesinvolved in the merger. Such measurements require knowledge of thegravitationalwaveforms predicted by GR for all relevant masses and spins. Numericalrelativity (NR) simulations are now sufficiently robust that we canaccuratelysimulate binary black hole mergers and obtain the waveform for all but themostextreme parameters, but they are too computationally expensive for a densecoverage of the parameter space. The effective-one-body model andphenomenological waveform models agree well with NR for the parameters ofGW150914, but could be insufficiently accurate for estimating theparameters ofa loud gravitational wave detection in other regions of the parameterspace. NRsurrogate models attempt to rapidly and accurately interpolate the waveformsfrom a set of NR simulations over a subset of parameter space. Using theSpectral Einstein Code (SpEC), we have built NR surrogate models for non-spinning binaries with mass ratios up to 10, and for spinning precessingbinaries with a restricted spin direction on the smaller black hole. Theytypically perform an order of magnitude better than other waveform modelswhencompared to NR waveforms which were not included in the surrogate trainingset,and can be used in gravitational wave parameter estimation.