| ... | ... | @@ -13,6 +13,79 @@ Leo |
|
|
|
Prat
|
|
|
|
|
|
|
|
Char / Minute Taker / Focus Session [Rota](https://git.ligo.org/groups/gstlal/-/wikis/West-call/Rota-table)
|
|
|
|
- Announcements (5 minutes)
|
|
|
|
- Please check the rota for next week's call Divya will be chairing, and Shio is the minute taker.
|
|
|
|
- Confirmation of next week's focus session
|
|
|
|
- None yet.
|
|
|
|
- Last week's East call https://wiki.ligo.org/CBC/Searches/GstLALEastAgenda20230629
|
|
|
|
- Prathamesh: Pitched the changes regarding to bankchi count tracker. Kipp didn’t like the idea and need to discuss with Chad.
|
|
|
|
- Quick updates (45 minutes)
|
|
|
|
- Operations (5 minutes)
|
|
|
|
- LL CBC operations
|
|
|
|
- Becca: We had several BBHs. Low frequency noise in Hanford, had our first retraction. Some known issues in GraceDb which didn’t bother us too much luckily. Only had some failures for super subthreshold candidates. Today we are flooded with Nagios alerts which Becca is trying to fix.
|
|
|
|
- Rachael: CIT maintenance still ongoing.
|
|
|
|
- Becca: Maybe we should not relaunch during maintenance as things run haywire.
|
|
|
|
- Leo: Did we remove counts for retracted event.
|
|
|
|
- Becca: No
|
|
|
|
- Prathamesh: During policy making we did not think about this. DQ people and we want to be conservative opposite direction. DQ people doesn’t put DQISSUES data label if they suspect there might be a signal. We only remove counts if we are sure, it is a signal.
|
|
|
|
- Becca: We should not remove counts for retractions.
|
|
|
|
- Prathamesh: We should only remove counts if we are sure about the event.
|
|
|
|
- Becca: It’s better to stick to a far threshold. Looking at qscans is subjective and hard to explain in papers.
|
|
|
|
- Leo: An objective criterion like ADVOK
|
|
|
|
- Shomik: Detchar are fine with ADVOK but had some issues in Hanford for the first and Livingston in second.
|
|
|
|
- Rachael: Since there is no DQ flag it is complicated. Agree with Becca to give a FAR cutoff.
|
|
|
|
- Cody: Existence of glitches doesn’t imply there is no signal. The pipeline knows much more than what one gathers looking at the omega scan. A simple decision tree helps to understand biases if something creeps in. Try to make the bias easy to track. If this introduces a bias, we should share it to the collaboration. Looking something and vetoing it may not increase the VT.
|
|
|
|
- Leo: Agree to Cody. Possible consequence of removing counts where we are not supposed to may cause more retractions down the line. We need a policy to be written down. Maybe HIGH_SIG + ADVOK label and make it official.
|
|
|
|
- Cody: How first do we need to remove counts.
|
|
|
|
- Prathamesh: If it is before the next event in the same bin it is OK.
|
|
|
|
- **_Condition of count removal is HIGHSIGNIF + ADVOK and we should wait for ADVOK label to be applied before removing counts. Remove counts from all analyses._**
|
|
|
|
- Yun-Jing: Should we remove counts for EW and Charlie.
|
|
|
|
- Ryan: It doesn’t matter and should be removed. (gut feeling)
|
|
|
|
- Divya: Does fewer bins in All sky analysis mean it will only remove counts from those bins in SSM?
|
|
|
|
- Prathamesh: there is no mapping. All analysis tracks its own counts. The remove job removes counts from the analysis’ count tracker.
|
|
|
|
- Leo: A signal in All Sky may be noise for EW and it doesnot matter if counts are removed or not.
|
|
|
|
- Ryan: Counterargument. Still effect the background which may have issues in the analysis.
|
|
|
|
- Becca: Rewhitening workflow is fine barring a few bugs which she has. Is there an issue regarding that?
|
|
|
|
- Leo: there was an issue which she will reopen soon.
|
|
|
|
- **Action Item: Discuss whitening workflow.**
|
|
|
|
- LL IDQ operations
|
|
|
|
- Rachael: Finish review and then sending it around.
|
|
|
|
- O4 Dev (30 minutes)
|
|
|
|
- Low latency integrated testing and Monitoring
|
|
|
|
- Becca: Review for GWSTAT going well. Few feature requests which need some volunteers for dev work. If people are interested, we can start weekly calls soon to get more folks involved.
|
|
|
|
- Template bank
|
|
|
|
- Debnandini: Trying to push an IMBH offline dag. Had some data issues which Ron is helping mainly regarding Sci-Tokens. Started a backup analysis on CIT.
|
|
|
|
- Likelihood ratio, background and foreground sampling
|
|
|
|
- Leo: Over the weekend had some suspicious events. We need to make dtdphi plots to see the time and phase difference in detector pair. Working to automatically update these plots to GraceDB. Shomik is double checking and instructions are on GWSCI.
|
|
|
|
- Injection file format
|
|
|
|
- offline DAG
|
|
|
|
- Cody: During offline reranked Patrick and Cody found 3 bugs. 1. It would unnecessarily write an empty snapshot. 2. We discovered that when a pipeline is shutdown an EoS message is send out, but no checkpoint message. So, need to refilter those times as the snapshots during those times are not written. However we might have the diststat files (Should triple check this!!) 3. The LLOIDTracker lock does not effect the SEGMENTTracker lock. So the snapshot durations are different between the segment file and trigger and diststat files. Cody wrote a commit to force the durations of trigger and the diststat files to be same.
|
|
|
|
- Leo: The corner case which Cody mentioned about a detection when an analysis is taken that. If we miss that during one chunk of open-box do you get it back in the next open-box?
|
|
|
|
- Cody: Such a situation is unlikely…
|
|
|
|
- Leo: Need a dedicate discussion with Chad.
|
|
|
|
- **Action Item: Come back to this snapshoting issue when Chad is here.**
|
|
|
|
- Shio: Has a dag for small BBH analysis on ICDS. Can add Cody’s and Patrick’s patch in the build and rerun.
|
|
|
|
- DQ dev
|
|
|
|
-
|
|
|
|
- HM search
|
|
|
|
-
|
|
|
|
- Exploratory development (5 minutes)
|
|
|
|
-
|
|
|
|
- Misc projects
|
|
|
|
-
|
|
|
|
- Paper Updates
|
|
|
|
- O4 rank stat paper accepted by PhysRevD.
|
|
|
|
- Focus session
|
|
|
|
- Shomik on Config Rework
|
|
|
|
-
|
|
|
|
- Divya: Have you tried making dags with this?
|
|
|
|
- Shomik: Just checked that validate inspiral was working. Didn’t make a dag people.
|
|
|
|
- Leo: Good if thec ROTA people to use this to create the rewhitening DAG. Might be risky too.
|
|
|
|
- Debnandini: Is there any issue if we donot use the rework?
|
|
|
|
- Leo: No scientific improvement. The rework is to make things user friendly.
|
|
|
|
- Divya: ‘No need to run anything, but just generate DAGs to sanity check if this works for each step. You may be able to point the filter options to existing svd and so you don’t have to run those before filtering.’
|
|
|
|
- AOB
|
|
|
|
|
|
|
|
|
|
|
|
## Action Items for next week
|
|
|
|
|
| ... | ... | |
