Simulation metadata has negative value for number_of_orbits

[Cyberface]

Hi,

I noticed that the number_of_orbits in the metadata for the SXS:BBH:3683 simulation has a negative value.

Not sure if this is the best place to put this but hopefully it's helpful!

import sxs
df = sxs.load("simulations").dataframe
df[df['number_of_orbits'] < 0][['number_of_orbits']]
# SXS:BBH:3683 | -33.713913

Cheers!

[moble]

Thanks for reporting this, @Cyberface.

@markscheel I don't know how this number is computed, but a few points may be relevant:

1. There seem to be a couple gaps in the horizon data.
2. There is some weirdness in the waveform even at the outermost radius, at times that roughly correspond to the gaps in the horizon data.
3. While there are ~800M of data for horizon C, there does not appear to be a merger or ringdown in the waveform.

import sxs
import matplotlib.pyplot as plt

sim = sxs.load("SXS:BBH:3683", extrapolation="Outer")
horizons = sim.horizons

plt.plot(horizons.A.time, horizons.A.coord_center_inertial)
plt.plot(horizons.B.time, horizons.B.coord_center_inertial, ls="dashed")
plt.plot(horizons.C.time, horizons.C.coord_center_inertial, ls="dotted")
plt.legend(["$A_x$", "$A_y$", "$A_z$", "$B_x$", "$B_y$", "$B_z$", "$C_x$", "$C_y$", "$C_z$"])
plt.xlabel("Time ($M$)")

plt.semilogy(h.t, h.abs)
plt.xlabel("Time ($M$)")
plt.ylabel(r"$|h_{\ell,m}|$")

[markscheel]

This simulation was run in 2015 and uploaded to the catalog in 2019. I flagged it as public in 2024, along with many other simulations that we found in the catalog and were not yet flagged as public for some reason.

Nowadays, the process of uploading a simulation to the catalog does various sanity checks including looking for missing chunks of data. Most of those sanity checks were not present in 2019, so this run fell through the cracks. The missing chunks of data usually were the result of filesystem issues on the cluster (including partial filesystem purges by sysadmins), or human error (in 2019 some people still concatenated their data sets by hand).

I am flagging this simulation as deprecated. The missing chunks of data occur in all resolutions.

[Cyberface]

Me again 👋

I noticed that SXS:BBH:3684 is also a bit strange and doesn't seem to have a merger either?

import sxs
bbh = sxs.load("SXS:BBH:3684")
w = bbh.h
plt.plot(w.time, w[:, w.index(2,2)].real)

[Cyberface]

Also just noticed that SXS:BBH:3664 doesn't seem to have a merger either

[moble]

Thanks for pointing these out, Sebastian. We'll be going through all of these runs, actually looking at plots for every waveform, before our next release of the catalog. I did this for our previous paper, but neglected to do so for the little data release a few months ago that was kind of a test for our next big release. I just assumed we had worked out the kinks in our data archiving process...

[Cyberface]

No worries, sorry to be annoying! I really appriciate the work the SXS colab are doing and look forward to the next big release!

[moble]

sorry to be annoying

Oh, not at all. I really appreciate it. Thanks again!