1st November Simulations Meeting

Attending: debbie, simon, warren, dave, ajc, jim, cathy, abi, kem, lynne, tony

Actions

  • Cathy to send the new opsim schema to the mailer 
  • Francisco, will talk with ajc about moving to git (after V3.0 in place)
  • ajc will look at the open source versions of slalib and we will have a simulations discussion about transitioning
  • Srini will represent OpSim at the camera geometry rework

 

Actions from last week:

  • ajc will talk to Iain about moving confluence off the 8443 port
    • Iain will address this after FDR
  • Steve r will discuss with the camera team the alignment of simulations with their work (2 week timescale)
    • ajc will follow up with the status of this
  • Simon will update catalog development plan - DONE
  • John will update phosim plan (DM roadmap is at ls.st/ldm-240) 

 

DESC meeting (Dec 4-6 Pittsburgh)

  • Debbie, Warren, Simon planning to attend
  • Discussion will include science requirements for the cosmology catalogs and running phosim


DM camera geometry work (Simon)

  • DM needs a framework for describing sensors, their properties and positions, including how to iterate over the sensors 
    • 6 months for delivery of this
    • Plan is python except for a detector class (with coord transforms) and then detector class in C++
    • OPSIM: needed outside of opsim for post-processing of outputs of opsim, may be needed inside of opsim for dithering (a long term goal)
      • probably only need positions at the amplifier level
      • Srini and Steve R will work with Simon to make sure Opsim is represented
    • PHOSIM: 
      • John will represent phosim on the camera geom redevelopment

Following is a part of an email sent to members of the OpSim and Camera teams.  I am very interested to have conversations with anyone having a stake in the design of the camera geometry classes.  Further, please feel free to add requirements to the design page.

I (Simon Krughoff) am leading the DM effort to redesign the camera geometry framework.  From the DM perspective the camera framework is responsible for several things.
1. Providing iterators over all devices in the camera.  This allows logical flows like:
  for every device in my camera:
    do some work
2. Providing coordinate transforms to/from a set of defined coordinates systems (e.g. pixel, camera, sky).
3. Providing a container to hold important electronic information for calibration as well as how to assemble amp segments into a full chip size grid.
The work is currently in the design phase and I would very much like to get input from the camera team and the OpSim team on requirements that are not provided by DM.
The current design document is at:
https://dev.lsstcorp.org/trac/wiki/Winter2014/Design/CameraGeom
The summary is that we intend to take a hierarchical approach  where cameras contain rafts and rafts contain detectors.  The implementation will primarily be in python with the detector class implemented in C++ so that the coordinate transforms contained by the detectors can be written in C++ and accessed by algorithms in C++.  The Detector class will be available in python via SWIG wrappers.  We intend both the hierarchy and coordinate systems to be configurable.

Crosstalk 

  • Tony mentioned that modelling cross talk may be significant for opsim
    • magnitude of crosstalk correlates with readout time so simulations of the impact of increasing readout could be beneficial
    • better understanding of the crosstalk on time scale of 2 months

Slalib transition:

  • opsim and catsim depend on slalib
  • we should move to an open source variant (Dave M reminded people that an open source version was mention last year)
  • action to look into how we can swap out the current version

Work plan for the next 3 months.

OPSIM:

a) Validate and deliver Opsim 3.0   -  Dec 1, 2013

      - cadence validation runs (~12), will release a new reference run, expect 4-5% fewer observations (due to increase in downtime) otherwise similar

      - CCB to define a reference run 

      - Simon: expect to ingest it. Cathy: a new schema so need to be aware of that

b) Docs for OpSim review  - Dec 23, 2013

       - Simulations group to review the documents (plus input from systems engineering)

c) OpSim dry run - Jan XX, 2014  (will know better by Wednesday)

d) OpSim review: week of Feb 3, 2014

e) Deliver OpSim 3.0 cadence runs:   TBD

f) Finalize Comparison FrameWork Design ? 


 

CATSIM:

  • Complete framework updates to allow external catalogs from RDMSs as well as flat files – Nov. 10, 2013
    • Code review – scheduled for the week of Oct. 28
  • Validate galaxy distributions relative to the simulations requirements – Dec. 1, 2013
    • Generate all validation figures and metrics presented in the validation document using new framework – Nov. 10, 2013
    • Compare metrics to requirements – Nov. 15, 2013
    • Make any necessary corrections and circulate realized distributions – Nov. 30, 2013
  • Complete framework refactor for generation of fully realistic inputs to PhoSim (precise astrometric transforms, galactic extinction, variability) – Dec. 31, 2013
    • Re-implement astrometric transforms as python Mixins – Dec. 10, 2013
    • Re-implement photometric transforms as python Mixins – Dec. 20, 2013
    • Re-implement code to sample SFD dust maps as a python Mixin – Dec. 31, 2013
  • Implement unit tests for all aspects of the catalogs framework – Jan. 15, 2014
    • Implement basic unit test framework – Completed
    • Identify necessary unit tests – Jan. 1, 2014
    • Implement unit tests – Jan. 15, 2014
  • Complete documentation and expanded example code set – Jan. 31, 2014

Future work needed (from the review)

  • Variability model, sky model needs to be in place before June, morphologies 
  • slalib needs to be removed 


PHOSIM:

  • 7/13:   PhoSim v3.3 Release
  • 8/13-11/13:   PhoSim v3.4 Development (PhoSim Reference Document plan)
  • 11/13-12/13:   BNL, FDR, & DESC meetings & preparations
  • 12/13-1/13:   PhoSim v3.4 I&T
  • 2/13:   PhoSim v3.4 Release


 

  • No labels