Child pages
  • 2014-10-24 Meeting notes
Skip to end of metadata
Go to start of metadata

Date

 at 9:30am MST/PDT

Tucson Location:  LSST Conference Room

Videocom: IP 140.252.24.8 (Direct Dial)

Call 1-866-330-1200, Participant Number 518-2287#

Attendees

Srini, Kem, Peter, Cathy, Simon, Lynne

Agenda/Discussion

Michael Reuter status and plans

  • Starting Nov 3 - working remotely -
  • Nov 17 traveling to Seattle to work with Lynne, Peter, Cathy, Kem 
  • Dec 15 in Tucson (Andy will be here & Francisco - roughly the time of the scheduler / simulator refactor design)
  • May travel to CTIO to work with Francisco after that

Scheduling workshop

  • $$ available? George not in attendance today, so no confirmation on money available 
  • Dates not yet confirmed
  • Andy will be putting together a proposal to George
  • Srini’s Profs - the have been contacted with no response, so we are assuming they won't respond and/or are not available

Status of releasing Opsim v 3.2

  • Stash master status
  • merging new config files - Kem will work with Srini to add config files at 1pm today
  • releasing 3.2
  • installation of v3.2 (and palpy)—new installation instructions needed? what about docker?
  • Srini was not able to quickly add the DB agnostic code changes so not included in v3.2
  • Filter change needed, we can put in parameters but still need minor coding changes - deferred for now
  • "Universal"/WFD tag - 
    • The new config files will include a key value pair Universal=True. 
    • In post-processing (via the conversion tool) we need to populate wfd field in the Proposal table with information from Config Table key-value pair.
    • Do we alter the conversion tool now to make use of the changes (wfd)? Who makes these changes?
    • We won't be using the SQLite branch yet (even for Docker) because the wfd tag is not yet implemented (and actually it is not in MySQL either)
    • Documentation - are there installation instruction changes; or feature changes that need to be added to Sphinx?

Docker

  • executes simulation using SQLite, (doesn't need conversion tool), executes maf on SQLite file (containing the wfd tag implementation). 
  • Currently we'll need a different image for every set of parameters. - need 1000 images for 1000 sets of params; but we could set it up so that based on a key, the desired set of configs could be downloaded- probably worth doing.
  • How do we set up for distributing with Docker (all are Linux virtual machine images)?
  • Still need to decide where to put the output files from the run.
  • Will this include the db agnostic implementation? or sqlite branch? (MAF runs just on SQLite now)

Update on database agnostic opsim development

  • Srini did some preliminary work and assessment and found it is too involved to be completed for the v3.2 release.

MAF and opsim status

  • Cathy/Kem update on usability and workflow
    • Kem likes the latest version - in the Metrics tab, there is a button that says plot metrics (scroll up and down to have to press the Plot button) place button so that it is more conveniently located.
    • Filter location pinned on the page. There may be browser dependent issues, but Peter and Simon think that we can fix that. 
  • Portablility of analysis results
    • Kem - can I do a six panel plot extraction for a publication?  how could I easily export this? Peter says they could write a small tool like "pdf join" to collect the desired images together. (pypdf - toolkit for merging documents in python.) May also be able to use this feature to create an offline access format.
    • MAF formatting issues for exporting/analyzing runs (Lynne suggests creating a tool to export the necessary files to a local computer without large data files, to make a portable copy of a selection of simulations and outputs)
  • MAF and the communitry—has anybody heard from the community?  are they using it?
    • Peter/Lynne -  Lynne has gotten some questions from about 4 different groups. How do we accept feedback for the cadences? Lynne would like there to be an attitude of hearing all ideas. 
    • Tom Matheson - NOAO Coffee talk on Tues - telling community about the released version of MAF and using it to create metrics to evaluate cadences. Also need to evaluate the science to make sure it doesn't break the goals (of course a set of metrics is still being developed by us and may require changes in the SRD).

Review sprint 14 (limited time was available, so deferred to offline conversations)

  • which items to backlog, which to sprint 15, which done (TBD)
  • Items for sprint 15 (TBD)
    • new tier 1 runs using v3.2

We are considering re-evaluating the time of this meeting to make it more convenient for Chuck, George and Andy (among others) to attend.  Cathy will send out a doodle poll.

Adjourned on time

1 Comment

  1. I can write the install instructions if you would like as I went through them with the docker install. They are quite nice and fairly simple with the refactoring that removes slalib and adds clouds.sql etc to the repo. It would be good to move the instructions over to a non-noao site when we rewrite them.


    From the notes, it wasn't clear if the conversion tool and v3.2 were ready to go. Can you comment - including whether v3.2 will run mysql and then output sqllite via the conversion or what?


    For docker we don't need a different image for a different config you just set docker to use a given config file (i.e. you check it out or you pass it to the container). The current version is set up to run as an interactive session but this doesnt need to be the case. Once v3.2 is out I will make the docker image work with sqllite, the conversion tool, whichever just let me know the preferred path