related to the GWTC-4.0 catalog paper https://arxiv.org/pdf/2508.18082v2
While the notebook in the zenodo entry states there should be ~1300 events in the files, in reality there are 347. Of the 129 events in Table 1&2 on the paper, only 73 appear in the zenodo entry, and 56 are missing.
I think the hpIGWN-GWTC4p0-1a206db3d_721-SearchSummaryTable.hdf5 file, because I do find the trigger files in the individual pipeline subfolders.
I’ll leave a short code to reproduce the problem & what it prints in the comments.
import gwosc
from gwosc import api
from gwpy.table import EventTable
# get GWTC-4.0 from gwosc api
catalog = gwosc.api.fetch_catalog_json("GWTC-4.0")
gwtc4_events = [v["commonName"] for v in catalog["events"].values()]
# manually copy-paste the events in tables 1&2 from https://arxiv.org/pdf/2508.18082v2
with open("table1.txt", "r") as f:
table1 = [line.strip() for line in f.readlines()]
with open("table2.txt", "r") as f:
table2 = [line.strip() for line in f.readlines()]
tables = sorted(table1 + table2)
# see the tables match the gwosc.api catalog
print(
f"{len(tables)=}, \n{len(gwtc4_events)=}, \n{len(set(tables).intersection(set(gwtc4_events)))=}"
)
# load the Search Summary from http://doi.org/10.5281/zenodo.17014083
Summary_Path = "candidate_data_release/search_results/IGWN-GWTC4p0-1a206db3d_721-SearchSummaryTable.hdf5"
events_table = EventTable.read(Summary_Path)
search_summary_events = list(events_table["gw_name"])
print(f"{len(search_summary_events)=}")
common_events = sorted(list(set(gwtc4_events).intersection(set(search_summary_events))))
print(f"{len(common_events)=}")
missing_events = list(set(gwtc4_events).difference(set(search_summary_events)))
print(f"{len(missing_events)} GWTC-4.0 events missing from zenodo entry: ")
for ev in missing_events:
print(ev)
to get the full set of 1382 candidates. If you don’t include that, then you get the first table in the HDF5 file, which happens to be the one for the minimally modeled cWB search, which is only sensitive to certain signals (generally higher-mass binary black holes–the cWB search looks for excess power without using a waveform model, so it is not so sensitive to long waveforms where the power is spread out in time).
If I make that change, then I get the expected results of
len(search_summary_events)=1382 len(common_events)=129 0 GWTC-4.0 events missing from zenodo entry: