Binary black-hole surrogate waveform catalog

This repository contains all publicly available numerical relativity surrogate data for waveforms produced by the Spectral Einstein Code. The base method for building surrogate models can be found in Field et al., PRX 4, 031006 (2014).

Several numerical relativity surrogate models are currently available in this catalog:

1. SpEC_q1_10_NoSpin_nu5thDegPoly_exclude_2_0.h5 — A surrogate model for binary black hole mergers with non-spinning black holes. This is described in Blackman et al., PRL 115, 121102 (2015). It is evaluated with the gwsurrogate python package, which can be found on PyPI. Instructions for evaluating this surrogate can be found in tutorials included with the gwsurrogate package and in this example IPython code.

2. NRSur4d2s_FDROM_grid12.h5 and NRSur4d2s_TDROM_grid12.h5 — These are fast frequency-domain and time-domain (respectively) surrogate models for binary black hole mergers where the black holes may be spinning, but the spins are restricted to a parameter subspace which includes some but not all precessing configurations. NRSur4d2s_FDROM_grid12.h5 is the NRSur4d2s_FDROM model described in Blackman et al., PRD 95, 104023, (2017), and NRSur4d2s_TDROM_grid12.h5 is built from the underlying (slower) NRSur4d2s time-domain model in the same way but without the FFTs. These surrogates are also evaluated using gwsurrogate, and a tutorial can be found in this example IPython code.

3. NRSur7dq2.h5 — This is a surrogate model for binary black hole mergers with generic spins. A paper describing it can be found at Blackman et al., PRD 96, 024058 (2017). This surrogate is evaluated through a standalone python package contained in NRSur7dq2.tar.gz, which has simple installation instructions in its README file. A tutorial can be found for evaluating this surrogate in this example IPython code.

4. NRHybSur3dq8.h5 — This is a surrogate model for binary black hole systems with generic mass ratios but restricted to nonprecessing spins. Before constructing the surrogate, the NR waveforms are hybridized with post-Newtonian waveforms to include the early inspiral. Therefore this model covers the full stellar mass range for ground-based detectors. A paper describing it can be found at Varma et al., PRD 99, 064045 (2019).  It is evaluated with the gwsurrogate Python package, which can be found on PyPI. Instructions for evaluating this surrogate can be found this example IPython code.

If you find these surrogate models useful in your own research please cite the Field et al., PRX (2014) paper as well as the relevant paper describing the specific numerical relativity surrogate model, if available (e.g., the Blackman et al. 2015 paper for non-spinning binary black hole coalescences).

Caveats:

1. Evaluating surrogate models outside of the ranges they were trained upon may give inaccurate results. Please use with caution when extrapolating.

2. The surrogate data available here for non-spinning binary black holes produced in Blackman et al. 2015 contains the (2,0) mode. However, this mode was not used in the paper. While this surrogate can predict a (2,0) mode, current numerical relativity simulations may not yet be able to accumulate (non-oscillatory) Christodoulou memory sufficiently. The surrogate (2,0) mode is founded upon basis SpEC waveforms that have been hybridized with leading order post-Newtonian waveforms. Therefore, the (2,0) mode can be included in the mode’s output but should be used with caution. Currently, the default option to evaluate this surrogate (using GWSurrogate) is to exclude all m=0 modes.