This repository contains all publicly available SpEC waveforms. All waveforms are provided at several fixed extraction radii, as well as extrapolated to future null infinity with different extrapolation orders (N=2,3,4), and corrected for extraneous translations and boosts of the center of mass. For many runs, waveforms are provided at several numerical resolutions (labeled, e.g., Lev1, Lev2, etc. Lev2 is higher resolution than Lev1).
Numerical accuracy can be assessed from the difference between different numerical resolutions. For runs with only a single resolution, look at a similar run. (Precession does not change numerical accuracy of our runs, but different mass ratios and spin magnitudes do).
Accuracy of extrapolation to future null infinity can be assessed by comparing different extrapolation orders. The 'best' extrapolation order to use depends on the simulation and what the waveform is being used for. Higher order tends to do better in the inspiral but worse in the ringdown. It is likely that the "OutermostExtraction" data is better than extrapolated data during ringdown.
For many runs, numerical resolution is set by changing Adaptive Mesh Refinement (AMR) parameters. AMR can make grid changes at different times for different resolutions, so occasionally comparisons of different resolutions might show unusual behavior different from the obvious convergence that would be obtained if comparing two different fixed grid sizes.
Numbers in our resolution labels 'Lev1', 'Lev2', etc. are simply labels, and are not necessarily meaningful when comparing different runs. That is, 'Lev1' of simulation A is not necessarily comparable accuracy to 'Lev1' of simulation B, if simulations A and B have very different mass ratios, spin magnitudes, number of orbits, or formulation of initial data.
Finite-radius waveforms contain near-field and gauge effects. They are included just as a check. Please use extrapolated waveforms instead, or the "OutermostExtraction" data, which has had some gauge effects removed and has less near-field contamination.
The initial data from many systems include large extraneous translations and boosts, which cause significant mixing between modes in the waveform data — even in extrapolated and CCE data. In all cases, waveforms should be taken from files with names including "CoM" (unless direct comparisons to older data are required).
Extrapolated waveforms work well for non-junk radiation and for non-memory modes (m != 0). If your work requires accurate junk radiation or m=0 modes, please contact us; we may be able to compute Cauchy Characteristic Extraction waveforms, which will not have these problems.
We include all (l,m) modes through l=8. The accuracy of small-amplitude modes will be less than that of dominant modes. In some cases (e.g. late ringdown, or when symmetries suppress certain modes), some (l,m) modes may be so small that they are purely numerical noise. Do not trust the accuracy of any modes smaller than ~1e-5 times the dominant mode.
The data in this catalog are free for anyone to use, but we request that you please acknowledge the first scientific publication of each simulation. These publications are listed in the file "metadata.txt" under the key "simulation-bibtex-keys". For your convenience, we provide bibliographical information in the form of a bibtex file here.
If you are interested in a joint project using this catalog, we welcome proposals for collaboration. If you have questions related to any particular data set or if you would like us to produce additional waveforms, please contact us and someone will get back to you.
The SXS Collaboration
This catalog has been made possible through the generous support of the Sherman Fairchild Foundation; NSERC of Canada, the Canada Chairs Program, and the Canadian Institute for Advanced Research; NSF grants PHY-0969111 and PHY-1005426 at Cornell, NSF grants PHY-1068881, PHY-1005655, and DMS-1065438 at Caltech, and NSF grant PHY-1307489 at Cal State Fullerton.
Simulations in this catalog were computed with the Spectral Einstein Code. Computations were performed on the Zwicky cluster at Caltech, which is supported by the Sherman Fairchild Foundation and by NSF award PHY-0960291; on the NSF XSEDE network under grant TG-PHY990007N; on the Orca cluster supported by Cal State Fullerton; and on the GPC supercomputer at the SciNet HPC Consortium. SciNet is funded by: the Canada Foundation for Innovation under the auspices of Compute Canada; the Government of Ontario; Ontario Research Fund– Research Excellence; and the University of Toronto.