West-call/101723.md

@@ -10,7 +10,7 @@ Divya
 
 ## Minute Taker
 
-Victoria
+Shio
 
 Char / Minute Taker / Focus Session [Rota](https://git.ligo.org/groups/gstlal/-/wikis/West-call/Rota-table)
 
@@ -21,8 +21,10 @@ Char / Minute Taker / Focus Session [Rota](https://git.ligo.org/groups/gstlal/-/
 ## Last week's call
 * [ ] Pratyusava, Cort, Divya: Discuss Condor upgrades at CIT
   - Divya: Not done yet
-* [ ] Debnandini: communicate that we’re doing an IMBH search to the CBC chairs
-  - Debnandini: For next week, I'll have a presentation at the CBC call and circulate slides beforehand.
+  - Pratyusava: didn’t get a chance to talk about it 
+* [x] Debnandini: communicate that we’re doing an IMBH search to the CBC chairs
+  - Leo: there was a question about how to combine the all sky search results and the imbh search results. It’s sth we’d wanna come back later. Surabhi opened an issue for that. 
+* [ ] Leo: snap shot bug fix 
 
 ## Agenda / Minutes
 
@@ -31,27 +33,73 @@ Char / Minute Taker / Focus Session [Rota](https://git.ligo.org/groups/gstlal/-/
   - Confirmation of next week's focus session
   - Do we want to restructure the West call?
   - Please double-check what accounting tags you're using. Make sure you're using online/offline and production/dev correctly
+  - O4a rota: https://git.ligo.org/gstlal/projects/-/wikis/O4a-GstLAL-Analysis-ROTA  
 
 * Last week's East call
 * Quick updates (45 minutes)
   - Operations (5 minutes)
     - LL CBC operations
+      - Rachael: does the rewhitening container have the correct accounting tag? -> I’ll check the repo
+      - Rachael: we had a bunch of issues. LLO internet connection shut off last week so data distribution data droping happened from Friday to Monday. One of the nodes on ICDS broke so we had to kill those jobs. These were caught by Icinga alerts. Some jobs on Edward and EW were hung up on http connections. Stuart et al had done work on this including hardware work to fix this but maybe it’s not fully fixed. May ping Stuart again at the end of the day if Stuart hasn’t responded yet. 1 BBH event. 
+      - Divya: Charlie’s AC length (701), SSM is a smaller bank and should have been using smaller (?) AC length. It was affecting the AC chisq values for all triggers, which would affect the LR and FAR assignments. I think we need to discuss more about this. 
+      - Leo: Chad and I talked about it last scrum. Chad assumes the chisq signal model may not be best proxy for the noise model. I might work out some math on how different the signal and noise model are. Chad thinks it should depend on bandwidth but nothing else. The KDE should be larger than what we’re using rn. 
+      - Divya: from the variance we expect (it should be higher?)
+      - Leo: the affect of the variance and the affect of the signal model can contribute differently 
     - LL IDQ operations
+      - Rachael: everything is fine. We can take this off the list. 
   - O4 Dev (30 minutes)
     - MDC for O4b prep
+      - Shomik: getting snr optimizer error fixed and getting NEMO to run an analysis. 
+      - Divya: analysis name discussion
+        - Shomik: it is a technically new analysis with new features and the statistics of the new analysis will be different. In the first week, there were 3 Jacobs and it was confusing from ops perspective. 
+        - Prathamesh: this is the first time where multiple things going on with the same analysis: MDC, reranks, online, etc. etc. I think it’s easier to have a different name for O4b analyses. Same for MDC. With a different name, we can identify easily which MDC is O4a and O4b. The MDC launched on Saturday is named Rick but no extra work required. Also makes us care more when its names we came up with. 
+        - Rachael: some concern from non-O4b ppl bc of confusion it can cause among non-gstlal ppl. It has long-lasting impact. It was already difficult to train ppl outside of this group with the scheme we had. Just being able to tell the diff between MDC and online is easy bc of the tags. You can do sth similar with O4b. O3a and O3b, the names were consistent bc the banks were the same. If you change the name from O4a and O4b, it can be confusing looking back. 
+        - Prathamesh: to that point, idk who outside the group need to know the names. Allocation names can be kept the same. But run the analysis under diff names. 
+        - Rachael: that will be confusing if you bring it up on CBC/AllSky calls. You may need to change allocation requirements etc. (Allocation names can be changed.)
+        - Prathamesh: working 1 week operating with Jacob when there were 3 Jacobs was already confusing. We want to prioritize our convenience over other’s convenience because operating under the same name is hard. 
+        - Leo: Maybe I want to see if o4b team means to keep Edward and Jacob names for production. Or will they change too?
+        - Prathamesh: haven’t decided on that. Preference is to change name for production. We can see how easy it is to work with new names. 
+        - Leo: I’ll be with Rachael. Internal usage, whatever name is good. But for production, we should be consistent. 
+        - Divya: o4b is starting after much shorter commissioning break. The toll it takes to even changing allocation name can be taxing. for consistency, it’ll be hard to change names within O4. Maybe better to not hastily change names although working for one week is hard. 
+        - Prathamesh: we don’t have to change allocation names to be clear. 
+        - Surabhi:  don’t know if we are counting votes, but I vote for not changing names 
+        - Patrick: agreed, that's gonna be hard to track
+        - Divya: continue using the names. Plz keep in mind the points that are raised. We can revisit this again if it is still a hassle. 
     - Low latency integrated testing and Monitoring
+      - Shomik: test suite is launched. Will work on dev work. 
     - Likelihood ratio and background sampling
+      - Leo: Prathamesh and I have been doing dev/fix for the offline workflow to integrate combined chisq signal model. Surabhi, have you tested the background collection patch? 
+      - Surabhi: the first one I tested didn’t have min instruments implemented correctly. The new one fails out. Haven’t looked deeply into the issues. Didn’t even run, crashed. 
+      - Leo: I tested but didn’t fail. Lets discuss offline. 
     - Template bank
+      - Debnandini: trying to run offline tests. Debugging. Presented the offline IMBH search on the CBC call. Issue raised by Tom Dent about how IMBH triggers will be included in the catalog paper and how it’ll be combined with all sky bank triggers. 
+      - Divya: I can help with segmentation errors 
     - Injection file format
-    - offline DAG
+      - Victoria: https://git.ligo.org/gstlal/projects/-/issues/811 This is sth we’re debugging for the new sim inspiral method and old sim inspiral method. Interesting results. Normal GW strain is larger than 10e-22. Two waveforms match for amplitude > 10e-22 looks similar but amplitude < 10e-23 does not agree. Is this diff due to new interface of LAL?
+      - Jolien: shouldn’t be. Goes through the same code. Not sure what’s causing this. Looks like samples are at the same times. Looks like there’s a slight phase shift. New interface just fills in info to the same code. Important to figure out the call stack, e.g. are these conditioned waveforms or not conditioned waveforms and what portion is causing this mismatch.
+      - Victoria: shouldn’t phase shift seen in all order of mag., instead of just low order of mag.?
+      - Jolien: not sure. Are you reading from xml files? 
+      - Victoria: yes. 
+      - Jolien: eliminate that. Just try the new interface. If it’s in routines that read text files, if the params fed to the waveform generator differs, that could be the origin of diff. Diff in conversion from text file to interface. 
+      - Jolien: just generate params, not waveforms, and check that they are the same. Masses and coal. phase and check that those values are the same up to double precision.  
+    - offline DAG/Ops
+      - Divya: launching new dag later this week to test errors that Debnandini and Victoria are seeing. 
+      - Debnandini: the run Victoria was using, are you using O4 data or O3 data?
+      - Victoria: O4 data 
+      - Divya: I’ve only run tests with O3 data so I’ll test with O4 data to reproduce the errors you’re seeing 
     - DQ dev
     - HM search
   - Exploratory development (5 minutes)
+    - Divya: any progress on the snap shot bug fix Leo took over from Cody?
+    - Leo: Cody DMed me, haven’t had a chance to work on it. 
+    - Divya: ping me for help. 
   - Misc projects
   - Paper Updates 
+    - count tracker paper published 
+
 
 * Focus session
-  - Review call
+  - Review call (recorded) SNR optimizer 
 
 * AOB