| ... | ... | @@ -10,4 +10,56 @@ |
|
|
|
* What are people interested in?
|
|
|
|
* Next meeting
|
|
|
|
|
|
|
|
# Recording and chat
|
|
|
|
|
|
|
|
**Note: Unfortunately my zoom dropped out twice during the call due to internet issues, so there are two gaps in the recording.**
|
|
|
|
|
|
|
|
https://dcc.ligo.org/G2100205
|
|
|
|
|
|
|
|
# Minutes
|
|
|
|
|
|
|
|
Attendance: Andrew Matas, Joe, Sylvia, Sanjit, Anirban, Jishnu, Andrew Miller, Yue, Huaike, Federico, Carole, Kamiel, Kevin, Deepali, Alba, (... apologies if I missed anyone!...)
|
|
|
|
|
|
|
|
Andrew went through slides in the agenda.
|
|
|
|
|
|
|
|
Sanjit: We should make code modular.
|
|
|
|
Andrew: Agreed.
|
|
|
|
|
|
|
|
Sanjit: It would be good to have an option to make dq cuts in pre-processing. Ideally we will have one set of dq cuts that all searches used, that is independent of the analysis being run.
|
|
|
|
Andrew: Yes, but our group won’t take responsibility for developing new cuts. However we can try to make it easy for people who will develop this.
|
|
|
|
Sanjit: Exactly. Just having a flag where DQ cuts can be specified in pre-processing will enable people to study new dq cuts later on.
|
|
|
|
|
|
|
|
Kamiel: Should we be worried about remaining 0.1%-0.5% difference between stochastic.m and stochastic_lite?
|
|
|
|
Andrew: The difference is below what we would report in a paper, so I don’t think it is urgent. But we certainly can dig into this and see if there is something to understand if there is interest.
|
|
|
|
|
|
|
|
Andrew went through “high priority” issues
|
|
|
|
These are issues meant to get us to the point where:
|
|
|
|
1. We can reproduce results from stochastic.m on a short stretch of data (eg without using condor)
|
|
|
|
2. We can do an injection and recover the results.
|
|
|
|
|
|
|
|
Sylvia: How independent do we want this to be? Bilby’s detector class does many of the operations like PSDs and filtering.
|
|
|
|
Andrew: I think we should leverage existing software. The conda environments make it easy for users to install LVK software.
|
|
|
|
|
|
|
|
Joe: It would be useful to have a roadmap.
|
|
|
|
Andrew: Agreed. We sort of have this information spread out over different places:
|
|
|
|
- flow chart of stochastic.m
|
|
|
|
- issues labeled high priority which are needed to reproduce results
|
|
|
|
- List of existing modules on main wiki page
|
|
|
|
But having a flowchart/roadmap is a good idea. I will make one.
|
|
|
|
|
|
|
|
Sylvia: Two pieces of advice from bilby experience:
|
|
|
|
1. Preserve option for ASCII output (binary output files can break if, eg, python versions change)
|
|
|
|
2. Have a contribution guide.
|
|
|
|
|
|
|
|
Andrew: I will make a contribution guide and will work on permissions. We might need to change something in the repository since at the moment the code lives in the stochastic project where I made everyone a maintainer. I hadn’t appreciated the point about ASCII output, but this is a good idea and we should definitely take advantage of knowledge gained from developing bilby.
|
|
|
|
|
|
|
|
Joe: How do we know what tools are already available?
|
|
|
|
Sylvia: Use issues as a way to communicate. In discussions here we can make
|
|
|
|
|
|
|
|
|
|
|
|
Anirban: I can give a presentation on how we implemented things in PyStoch.
|
|
|
|
Andrew: Great idea. We also don’t need to wait to discuss this, we can discuss on mattermost or issues or email.
|
|
|
|
|
|
|
|
Action Items
|
|
|
|
* Andrew will make a contributing guide, roadmap, mattermost channel, and set permissions so there is a protected master branch.
|
|
|
|
* Everyone interested will take a look at existing code and issues, feel free to make suggestions or new issues, comment on existing issues, and/or sign up to work on existing issues. However, please avoid making changes until Andrew sets up permissions and a contribution guide. |
|
|
\ No newline at end of file |
