Action items from last week: https://git.ligo.org/groups/gstlal/-/wikis/West-call/020822#action-items
* Last week's East call
update on svd bank from Victoria. The solution would be manage column issues in sim insinrpal. Chad will bring this up in this weeks’ east call.
Quick updates (45 minutes)
* Operations (5 minutes)
* LL CBC operations relaucnhed bbh analysis after fix the bug. Still waiting for the likelihood to be computed.
Mario : Divya relaunch Mario with the Vicrotia;s max duraiton bank. It seems working fine. Still seeing issue the data dropping.
Chad : Data drop is due to the data falling behind.
Alex : some library is unoptimized. gstreamer build is optimized now and so rebuild container.
Chad : eys, issue is about unoptimized C code being run in the analysis and profiling suggests the jobs are running ~20% slower due to the unoptimized build.
* Offline CBC operations
Looking into VT estimate of O3b SSM
* LL IDQ operations
N/A
* O4 Dev (30 minutes)
* Low latency integrated testing and Monitoring
Becca : I want send_inj_stream to handle different config (e.g. loading inejctions from disk or having them in the MDC).
* Template bank Chad :
Shio and Chad met to chat about the plan to make a new bank (fewer temapltes than O3) and Divya will decompose it with Victroia’s patch. and run analysis on the bank to get
* Likelihood ratio, background and foreground sampling
update in two weeks about time-dependent background
* Optimization and throughput benchmarking
Alex : touch based on some profiling in LL analysis
* Data format wrangling
Chad : started down the road of revisit table design. Somethign to do with injection table. The code has to be reworked to compatble with the new input and modify the siminspiral table.
* Enabling running on OSG/IGWN grid
Patrick : Stuart made proposal condor submit . At some pint future when you submit job in condor. It will crash if you don’t require disk space. It would be good to do ahead of time before submitting dag.
* DQ dev
N/A
* Exploratory development (5 minutes)
N/A
* Misc projects
Focus session
* iDQ Dev (Rachael)
Chad : For slide4 I like to understand if this is doing what we want it to be doing. Should do some test like ROC curve. The majority of timeseries is at zero. The number of chunks higher than zero need to be checked and see if that is consistent with the number of glitches. Having self-consistenct test results would be convincing.
Becca : what kind of metrics will be displayed in the idq dashboard? Rachael : time series? Of idq terms in likelihood. There should be a link in the dashboard to the idq board too.
AOB:
Stefano : some weird inejction revcovery
Chad : tbh, no idea. What needs to be done rerun the O3 chunks with master after having git hash. Use ICDS
Stefano : I can take that.
Chad : have all the recent infrstracture. If that is all taken care, we can basically pick whichever chunk.
Ryan : Ive set up some config in CIT but didn’t get through but can help with in ICDS.
Chad : we will need to do many of this towards O4 anyway.
## Chat log
<10:01:06> "Becca Ewing": oh right
<10:01:53> "chad.hanna": hand up
<10:04:29> "Leo Tsukada": hand up
<10:06:20> "Becca Ewing": i think we are
<10:08:12> "Leo Tsukada": ah, detector response would be different over time...
<10:09:10> "SurabhiSachdev": bayestar can also simulate snr time series
<10:09:26> "chad.hanna": yeah, lets let bayestart just simulate it
<10:09:39> "Leo Tsukada": @Surabhi : could you point us to place to look for that?
<10:10:01> "Shio.Sakon": @chad i posted the number of templates (as of now) in the O4 temp bank mattermost channel
<10:10:29> "Shio.Sakon": nbank = 400 (7 more jobs to be done): 104061
nbank = 50 (6 more jobs to be done): 45257
(the numbers of templates are drastically different for the two settings of nbank. Is this normal or should I be concerned?)
