- Divya: Prat is not able to finish the testing, needs to test in small analysis
- Prathamesh: Move this task to sprint?
- Cody: I disagree, task to people outside penn state should be on this call
* [ ] Leo: snap shot bug fix
- Leo: still ongoing
- Not yet reviewed
* [x] O4b Team: Gantt Charts
- Saw last time
- Shomik: more information about analysis names
## Agenda / Minutes
* Announcements (5 minutes)
- Please check the rota for next week's call
- Confirmation of next week's focus session
- We have one today
- move to a system where each week there is an in-depth discussion about someone’s project
- Feedback on the restructured west dev call
- Try small changes first, make more changes along the way
- Feedback are welcome
* Last week's East call - no east call
* Quick updates (45 minutes)
- Operations (5 minutes)
- LL CBC operations
- Becca: no updates
- Becca: Well actually we are moving gwistat to a kubernetes deployment today, that’s my only LL monitoring update
- KAGRA discussion: see [issue](https://git.ligo.org/cbc/action_items/-/issues/53)
- Make sense to keep this here until we make a concrete decision about what we do for KAGRA
- O4 Dev (30 minutes)
- Low latency integrated testing and Monitoring
- no update
- O4b online production lead team updates
- Shomik: we have not checked the 3rd event
- Shio: we found the event, was not able to remove counts
- Shomik: Reranks are in progress, should have kafka servers next week
- Chad to shio: what are the errors?
- Prathamesh: will investigate, internal server error
- Chad: Web server failure, python traceback? Check through bottle? Probably there is legit bugs in corner cases
- Prathamesh: There has been event plotter errors that cannot pass through coincs
- Leo: Have people launched EW MDC?
- Amanda: not yet, we met early this morning to fix config, there is an error trying to generate makefile, Alvin made a fix, hopefully we will launch by tomorrow.
- Leo: Curious about background collection bug fix
- Prathamesh: Were you able to get o4b-ew branch set up?
- Amanda: yes
- Prathamesh: O4b mdc have not been using bank-chisq as we intended, but it has been collecting bank-chisq, so we can check through rerank.
- GWTC (allsky, IMBH)
- Chad: meeting with ___ try to figure out how to use catalog tools. People more or less feel good about current gracedb schema. Hope to ask Alex to move it to production.
- Chad: Offline injection/CVFMS drama. Help Divya with git issue. Discuss during dev/ops call tomorrow. Two pieces: CVFMS/ R&P group
- Cody: no update on R&P side to getting things before november.
- Chad: Suggest Josh to take old injections to test plumbing
- Cody: Thought Reed was main one. Do you know if Tom is involved?
- Chad: no, Tom was just the one who spoke up this morning
- Cody: Dag taking in injections, plotting jobs are hanging, will debug. Good time to test filtering, one smooth analysis
- Prathamesh: Plotting job super long?
- Cody: Running for 4 days, so killed them
- Prathamesh: Old version, injection clustering not working correctly.
- Cody: Can you @all when there are bugs found? Better communication?
- Prathamesh: Bug was solved months back. Git was doing something weird?
- Cody: can believe that
- Likelihood ratio, background and foreground sampling
- Leo: about the controversial event 231020bw, me and shomik looking into that. Try getting different SNR time series in H1, make dtdphi plot with different marker.
- Injection file format([issue](https://git.ligo.org/gstlal/projects/-/issues/822#note_830165 "remake waveforms of two methods for a boarder investigation on the discrepancy") and [issue](https://git.ligo.org/gstlal/projects/-/issues/811#note_830106 "Investigate the amplitude discrepancy between sim_inspiral waveforms and sim_cbc waveforms"))
- Victoria: Running small offline analysis to check injections. Last week mentioned maybe we should remove this item and merge this with the item with Rachael.
- Prathamesh: Another focus session?
- Chad: Yes, soonish. Migration path? What’s ready to merge? When the first round of changes can go in
- Victoria: Can give focus session next week
- offline DAG and Ops
- Prathamesh: Fixing bugs in offline workflow, lots of bugs with condor file transfer. Divya suggest an offline-day.
- Divya: would friday work? Get it done in next sprint. Fix offline bugs and features that has been slipping through the cracks
- DQ dev
- Rachael: can remove this
- HM search
- SNR optimizer
- Prathamesh: Zach work on quantifying how PSD improve SNRs. I worked on SNR phenomenological review, check the optimizer produced similar SNRs with gstlal. (Showed plot) Force optimizer to use same PSD as gstlal, and no template optimization, SNR is very similar to gstlal’s.
- Jolien: Why aren’t these identical?
- Prathamesh: 1. Recovery time is different. 2. PSD for data whitening in gstlal is not static. PSD we are using is one snapshot of the PSD at the end of the calculation. 3. gstlal’s SNR calculation is the reconstruction of svd templates.
- Jolien: Good to isolate these changes out. Use static PSD for gstlal.
- Chad: make sure segmentation is the same. Whitening filter is weird. lock to same whitening cadence.
- Jolien: Alarming things: 1. time offset, that’s potentially concerning. One point in time, slight issue with timing. Had to do with acausal whitening. Impact sky localization.
- Leo: is this event from o4? mdc?
- Prathamesh: random one of o4b mdc
- Leo: check sky map? New whitening kind of affected sky map, especially for HLV. Find 3 ifo events and check skymap. Look at search area.
- Jolien: timing discrepancy doesn’t tell you what’s right and what’s wrong.
- Leo: MDC, we know true location of source.
- Prathamesh: To solve this, look at injection and check if gstlal vs SNR optimizer recover correctly.
- Chad: Histogram of injection recovery.
- Chad: one sample offset? is same rate 2k Hz? (Prathamesh: Yes) Each tick: 40 points. There is a correction in gstlal to coalescence time. LAL defines an epoch and returns an offset. Subsameple interpolation mismatch?
- Jolien: Use impulse response to check?
- Chad: Jim will do that
- Zach: PSD contribution plot?
- Prathamesh: Show on thursday
- Computing
- update on the status event follow up script (Alvin)
- Alvin: update to make file of event followup script. Makefile to python version. demonstrated how to run event_followup.py. Automatically grab coinc. Different from previous: reading in process_params_table, no more hard coding cutting. Grab DQ segments file, problem: slashs in comments, need to remove slash. Problem: running gstlal_inspiral job in singularity: gracedb servers url error. Demonstrate that the code can generate the SNR time series and its plot.
- Prathamesh: check this into gstlal?
- Alvin: look into what frames should be used. frame dir only stores 300s data, suggestions to where to look for frames?
- Exploratory development (5 minutes)
- Misc projects
- Paper Updates
* Focus session
- Injection Streamer (Rachael)
- Rachael: SCCB ticket already approved a long time ago, can we deploy? Or make new SCCB ticket?
- Patrick: make new SCCB ticket anytime you make a change
