Maintenance will be performed on git.ligo.org, containers.ligo.org, and docs.ligo.org on Tuesday 29 April 2025 starting at approximately 9am PDT. It is expected to take around 30 minutes and there will be several periods of downtime throughout the maintenance. We will be switching the authentication mechanism to use the system that we will be using with the new instance, it will still be utilizing the Federation Proxy but the underlying communication method will be changing. From the user perspective this should be the same as it currently is and you should not notice any differences. However there may be issues that we have not anticipated nor encountered during testing. If you have any comments, questions, or concerns please add them to the helpdesk. In addition we will also be installing the latest security update.
One issue is that the clustering used in gstlal_inspiral internally in online
running is set by the parameter "thinca-interval", which we usually set to 1
second for latency reasons. That means that triggers are clustered over a 1
second interval. The offline clustering SQL file uses a ± 4s window.
So I modified that script to use a ± 0.5s window, which I hope will
approximate the gstlal_inspiral internal clustering well enough. See:
88cebe1656e267fd9d6579db83e4cdfff00c804e. This was not however sufficient to
fix the IFAR plot. I still think that this is simply due to the fact that there
has not been sufficient data to overwhelm the very conservative prior. Lets
see if the situation improves as we go forward.
- Money plot here looks strange (dominated by priors, insufficient data, ...); it would be good if release versions can fix this: <https://ldas-jobs.cgca.uwm.edu/~gstlalcbc/online-test/Images/H1L1V1-GSTLAL_INSPIRAL_PLOTSUMMARY_ALL_LLOID_COMBINED_05_count_vs_ifar_openbox-1121652724-550359.png>. Note, I think this is the expected behavior, lets wait to see how it evolves during the replay
\subsubsection day1completed Completed Actions
...
...
@@ -146,3 +157,22 @@ In attendance: Chad, Edward Maros, Gareth Tomas, Jolien Creighton, Kent Blackbur
- Make the information about injection rate available on the summary pages along side the parameters used for injections: Done see above
- Instead of overwriting online iHope pages it would be good to save the old one (with a time stamp) and start afresh but with all the data so that we can have the history available to look at. Done, see: 43cc68e42c882a4710196ebe51633432b0ad2bca
- Use a smoothing kernel with a larger (try 2, 4) scale to smoothen the likelihood and probability plots. This has been investigated and is documented here: https://bugs.ligo.org/redmine/issues/2339
\subsection day2 July 30, 2015
In attendance: Chad, Jolien and Sathya
\subsubsection day2actions Actions
- Look at missed close-by injections (there are not many but it is puzzling why ~8 Mpc injections are missed; could be due to injections being on lock-loss boundary but should be investigated).
- Use integer GPS second for the on-line clustering script.
- The SNR heat plot and the SNR-Chisquare trigger plots do not appear in SIM-DB page; it would be good to have them there too to compare with the Big Dog event and to see if the anatomy of the triggers is similar when there is a loud event.
