GW231123: a Binary Black Hole Merger with Total Mass 190-265 Msun

This data release contains the analysis results and data behind the figures of the GW231123 discovery paper. Strain data for this event (the H1:GDS-CALIB_STRAIN_CLEAN_AR, L1:GDS-CALIB_STRAIN_CLEAN_AR channels) can be downloaded on GWOSC (link). The content of this repository is:

Results: 

The H1_subtracted_frame.tar.gz  file contains 20s of strain data from LIGO Hanford from the channel: H1:GDS-CALIB_STRAIN_CLEAN_BAYESWAVE_S00. In this channel, a glitch at [1.7,1.]s before the GW event has been subtracted using BayesWave. This is the channels used to infer the properties of GW231123.

the posterior_samples.tar.gz file contains the combined posterior samples for the 5 different models described in the paper.  In addition to containing the posterior samples, the .h5 file also contain metadata about the analyses including the configuration files (which specify details such as the detector data analysed), noise power spectral densities and calibration uncertainty envelopes. The inference of the source parameters were performed with Bilby and we assume the calibration envelopes for LIGO Hanford and LIGO Livingston maps calibrated data to theoretical strain. The .h5 files are formatted using PESummary. 

The Additional_results.tar.gz file contains the data necessary to reproduce Fig.1: whitened data (BayesWave_whitened_data_H1.dat, BayesWave_whitened_data_L1.dat), BayesWave reconstruction (BayesWave_signal_median_time_domain_waveform_H1.dat, BayesWave_signal_median_time_domain_waveform_L1.dat), CoherentWaveBurst (cWB) reconstruction (cWB_H1_wf_signal.dat, cWB_L1_wf_signal.dat), median waveform and 90% uncertainty from combined samples for 5 waveform models (median_combined_waveform_with_90pct_uncertainty.dat, created from the script combine_whitened_waveforms.py). The median waveform with 90% uncertainty for each waveform model are included here as well. The mismatches to reproduce Fig 2 (MatchResults_HighSpin.csv, MatchResults_LowSpin.csv). The folder pyRing_posteriors contains the posterior samples for the ringdown analysis considering a damped-sinusoid model at late times (26,28,30M).

the Additional_posteriors_samples folder contains : combined posterior samples used for Fig. 8 (SXS_BBH_0483_posterior_samples.h5, SXS_BBH_4030_posterior_samples.h5), and the posterior samples using only LIGO Hanford and LIGO Livingston used in Fig.9. These files have been formatted in a similar way to the posterior_samples.tar.gz file.

The scripts.tar.gz folder contains (please note that for the following scripts the path of the input files as to be completed by the user):

notebook to reproduce Fig.1

notebook to compute mismatches and reproduce Fig.2

In the folder 'Figs_source_properties': 'production_plots.py' scripts to reproduce Fig.3, Fig.4, Fig.9. 'systematics_production_plots.py' to reproduce Fig. 7, Fig. 8, 'utils.py' script tp open the posterior samples

script to reproduce Fig.5

notebook to reproduce Fig.6

script to reproduce Fig.10

The figures.tar.gz folder contains the Figures presented in the discovery paper

Differences with respect to the previous version on Zenodo (v2)

Since the original submission, the PE samples have been updated following a change to the SNR calculation in the Bilby code (see arXiv:2508.11091), and to be consistent with the final GWTC-4 catalog results. This change has resulted in a lower reported SNR (median 20.7 versus the previous value of 22.6), but all other changes are minor, and none of the key results or conclusions of the paper are changed. Figures, table, and the relevant text and numbers throughout the paper have been updated accordingly.

The results presented in the published manuscript are based on this data release (v3).