wiki:DC3APostMortemMeetingScienceAppsI
Last modified 10 years ago Last modified on 08/31/2009 07:02:04 PM

DC3b scope decisions

  1. Multifit will be run for only a single model, the Small Galaxy (SG, in the nomenclature of the DM document.
  2. DC3b pipelines will be assembled early in the process, with empty stages as needed, so that we always have a fully runnable system.
  3. We will use the cadence from OpSim for simulated data being input to DC3b
  4. The atmosphere module from Jim Bartlett's group will be connected to ImSim to feed into image simulations for DC3b
  5. We are not dependent on CalSim for anything in DC3b. The CalSim group will attempt to provide some output data for us to test with by August.
  6. DC3b will not process crowded fields.
  7. In DC3b Data Release Production, we will "Pre-sequence Images" in known sky sequence once.
  8. We will not "Pre-sequence catalogs", assuming scalable architecture will be adequate.
  9. In ISR, we will not include cross-talk correction.
  10. We will generate pupil ghost images by hand for CFHT data. We will use existing fringe frames for CFHT. Sim data?? AC
  11. We will stub out a deblender in DC3b. We will reject or fit a bad model to things that look like multiple objects.
  12. We will do detection on multi-band chi-square image, then independently fit to multiple bands in deep detection.
  13. We will do monochromatic co-adds, within a filter, we need to reject outliers, may require 2 passes.
  14. We will take one field of DC3b data and create a catalog with grids of stars?? AC (KT special analysis tools?)
  15. We will not determine astrophysical sky background from stacks of images, will be done on image by image basis. We will persist what we get from Image characterization, included subtracted sky models.

Actions

  1. Need a cross-talk spec for the Camera. TA took the action to specify a level of cross talk below which it can be ignored. NOW OUT OF SCOPE FOR DC3B, NEEDED FOR R&D SAME TIME.
  2. In the context of RHL's alternate scheme for processing the two exposures of a visit, we need first to assess how well the current one is doing. This requires analysis of the catalogs resulting from good image subtractions. Serge noted that the policy file for the Association PL has a number of parameters that affect this, and that they are currently guesses. TA took the action to check them, possibly suggest better values. We also need real data, rather than the mocked up exposure pairs, to test this well. It's not clear that we have anything suitable. TA took the action to look for usable data source. DONE.
  3. SM to do run with cosmic ray detection turned off to see how it is doing.
  4. TA will check to see whether the PSF width of the template is less than that for the science image for the DC3a subtractions that are giving high spatial frequency noise in the subtraction footprints of bright objects. Now know that diffim is not deconvolving. DONE.
  5. KTL/JB to have writeup with test numbers for at least two or three scalings for database scalability. Prior to PDR. DONE.
  6. TA to contact the SS Collaboration re the type of orbital elements that they want in the catalog. DONE.
  7. LJ to work with SS Collaboration, answer by end of July, 2009.
  8. SAT to walk through entire DRP flow to identify outputs. In schema design we will identify persisted ones and map to schema, including defining units for all database columns, whether IAU name is in Object table schema, etc.
  9. We need a long-range timeline showing how we will be using various kinds of input data, simulated and real, to the DM system. NOT FOR DC3b, FOR PDR
  10. We need a real execution time model for DC3b. GPDF.
  11. TA to do quick mod to existing execution timeline to change image transfer time from mountain to base from 6sec to 1sec, and to include third exposure.
  12. We need per-ccd qe vs wavelength in the ImSim. Still need to specify data products in ImSim plans. DONE.
  13. TA to get to AJC a list of physics needed in ImSim to support DC3b. DONE.
  14. DM needs a general focal plane geometry model. SAT to review proposal.
  15. Lynne find CFHT camera, HSC focal plane documentation
  16. Andy Rassmussen to see if these can be modeled using his model in use in ImSim.
  17. TA to modify document on simulated inputs required for DC3b, including needs for multiple filters, and decide on number of images needed. (See DC3b\Data Inputs\Sim data). Expect 3 ImSim deliveries: initial small data set to ensure that physics are there, larger production for integration testing, and final full for performance testing.
  18. RHL to form a working group including at least UCDavis team and Mike Jarvis to discuss deep detection issues, especially generation of PSF's. DONE, AUGUST 17 - 18, 2009.
  19. J Bosch to contact James Jee should be part of this discussion. DONE.
  20. We need a wider range of SEDs for stars in the input catalog that generates the simulated images. Lynne working with Science Collaboration on this, adding white dwarfs, others. Lynne to check D. Monet requirements. Lynne to post link to catalog documentation to lsst-data. TA to sign off on distribution.
  21. J Bosch draft proposal for SAT and calibration experts suggesting different process flow for deep detection.
  22. TA ask Ryan Scranton create a proposal for DM-wide decision about how to tessellate the sky to SAT. URGENT

DC3b Task List

DC3b Task List for ISR (8/4/09)

  • Generate fringe frames and ghost images for subtraction (deferred)
  • Apply fringe frame correction (Krughoff)[3wk]
  • Develop device information classes (CCDInfo and AmpInfo?) (Krughoff) [2mo]
  • Develop Calibration Products Database class (Krughoff) [2mo]

DC3b Task List for Difference Imaging (7/29/09)

  • Adapt Psf spatial cell model for ip_diffim (Becker; #876)
  • Implement sum-of-gaussians (Alard-Lupton) kernel basis (Becker)
  • Investigate regularizing delta-function kernels (Becker)
  • Fix ticket #873 [2 wks] (Owen)
  • Investigate improving the spatial modeling of the kernel (Becker)

DC3b Task List for Association Pipeline (7/29/09)

  • Interslice communication (Monkewitz)
  • Post spacial matching (Monkewitz)
  • Run association pipeline without ISR cosmic ray (Monkewitz)

DC3b Task List for WCS (7/29/09)

DC3b Task List for Coadd Generation (8/4/09)

  • Integrate HealPix, our chosen sky pixel representation for DC3b, into our software stack; see http://dev.lsstcorp.org/trac/wiki/ImageAccessApis#Pixelization (This task may already be listed elsewhere). ? (?)
  • Code to warp images to/from our sky pixel representation [2 wks] (Owen)
  • Code to create PSF-matched difference imaging coadds (no outlier rejection) [3 wks] (Owen)
  • Create a set of difference imaging coadds ? (Axelrod?)
  • Code to measure quality of difference imaging coadds [2 wks] (Becker/Owen?)
  • Test difference imaging coadds (Becker)
  • Implement image access for coadds (KT) (listed separately at ImageAccess)

DC3b Task List for Deep Detection (8/4/09)

  • Code to create chi squared deep detection coadds (no outlier rejection): [2 wks] (Owen)
  • Create a set of chi squared deep detection coadds: ? (?)
  • Code to measure quality of deep detection coadds: ? (?)
  • Chi squared detection (Lupton)
  • PSF for deep detection (Jarvis)
  • Simplified background estimation (Lupton)

DC3b Task List for Source Measurement (7/29/09)

  • Stellar photometry (Bickerton)
  • Implement image access for multifit (KT) (listed separately at ImageAccess)
  • Multifit: (All times are design and coding only. Add 50% overhead for testing, reviews, etc.)
    1. Complete API for model fitting with marginalization over calibration parameters. [<1 wk] (Dubcovsky/Bosch?)
    2. Implement convolved Sersic models which can be evaluated and differentiated in different coordinate systems than their parameters are defined in.[2 wks] (Dubcovsky/Bosch?)
    3. Implement multifit fitter.[4wks](Dubcovsky/Bosch?)
    4. Implement multifit pipeline which leverages multifit API and image-access framework [1 wk](Dubcovsky)
    5. Implement pipeline stage for extracting ellipse parameters, flux, and bounding box from detections (footprints?). [ <1 wk : part of task 4](Dubcovsky)
    6. Implement constrained models for forced photometry [2wk](Dubcovsky/Bosch?)
    7. Create LSST stages for forced photometry. This is a large change to the pipeline and the data volume. However, the development of a stage to do forced photometry on single exposure of difference image is a simple task. [1wk](Dubcovsky/Bosch?)

DC3b Task List for Astrometry(7/29/09)

  • Define sky HEALPix size (Monet)
  • Define image position algorithm (Monet)
  • Determination of low order terms (Monet)

DC3b Task List for MOPS(7/29/09)

  • Integrate DayMOPS into LSST framework (Francesco, Myers)
    1. Write LSST Pipeline, Stages for findTracklets, collapseTracklets, linkTracklets and IOD, using old Auton libraries: Francesco, DONE.
    2. Integrate MOPS Stages with new C++ KD-Tree tools: Myers, [1wk]
  • Incorporate OpenOrb into MOPS IOD, ephemeris generation (Francesco)
    1. Develop Python bindings: DONE
    2. Integrate bindings into LSST Pipeline: ??
  • Identify cluster for LSST-MOPS Runs: Myers [1wk]
  • Run with Solar System Model: Lynne, Myers [1wk]
  • Identify catalog for real images: Descoped?
  • Recode KD-Tree apps in C++: Myers, Cleveland
    1. Rewrite findTracklets, unit tests: DONE
    2. Rewrite detectionProximity, fieldProximity, orbitProximity and unit tests: DONE
    3. Write linkTracklets and unit tests: [3 wks]
  • Identify SDQA Metrics: Francesco, Lynne, Myers
    1. Already done(?)
  • Deliver calibration catalogues: Burke
  • Investigate Op Sim issues: Lynne, Myers [2 wks]
  • Extended SSM for LSST: T. Grav [3 wks]

DC3b Simulated Data (8/4/09)

  • Catalog Level Simulation
    1. Generate catalogs directly from OpSim DB [4 wks] (Gibson)
    2. Generate catalogs based on user defined query [1 wk] (Gibson)
    3. Generate ugrizy magnitudes (magnitudes above atmosphere, at nominal airmass, at airmass of observation) using appropriate throughputs [2 wks] (Gibson)
    4. Implement atmosphere as given by Bartlett based on OpSim outputs [2 wks] (Gibson)
    5. Implement corrections for mean apparent place (stellar aberration etc) [3 wks] (Pizagno)
    6. Develop stellar catalogs with appropriate distribution of stellar types [4 wks] (Gibson/Jones/Juric?)
    7. Create equatorial stripe for stellar catalogs using Juric model [2 wks] (Gibson/Jones?)
    8. Implement variability for 10% of the stellar sample [2 wks] (Krughoff)
    9. Implement Galactic extinction model [3 wks] (Gibson)
    10. Create galaxy catalog with tiling to cover DC3b area [3 wks] (Pizagno/Connolly?)
    11. Implement 3 component galaxy model (2 sersic and point source) [3 wks] (Pizagno/Connolly?)
    12. Implement galaxy extinction for sources [2 wks] (Pizagno)
    13. Create moving object catalog at depth of PS1 [1 wk] (Connolly, Jones, Krughoff)
    14. Create optimized database for serving MOPS catalog data [3 wks] (Krughoff/Connolly?)
    15. Generate hot pixel and cosmic ray database [2 wks] (Connolly)
  • Image Level Simulation
    1. Inputs (coordinates and site/sky) as given by OpSim (Pizagno)
    2. Implement rotation of telescope, camera (Peterson)
    3. Implement simple control system model (jitter of tracing and rotation) (Peterson)
    4. Generate ugrizy magnitudes that match the QE from 1089 (Jernigan)
    5. Verify chromatic differential refraction (Pizagno)
    6. Match inner and outer psf (including scattering off surfaces, mie scattering) (Jernigan)
    7. Develop new atmospheric screens with expanded inner and outer scales and coherent across 2 15s exposures (Jernigan)
    8. Create structure function for atmosphere that matches the cerro pachon (Peterson/Jernigan?)
    9. Implement new focal plane model from camera team. (Peterson)
    10. Implement new model for the spiders (Peterson)
    11. Implement variable chip-to-chip QE (Jernigan)
    12. Verify sky background calculations (Jones/Peterson?)
    13. Apply CRs and hot pixels to the images (Connolly)
    14. Format image for DM (Pizagno)