Uploaded image for project: 'Data Management'
  1. Data Management
  2. DM-12773

Explore not using the SIP terms to initialize jointcal

    Details

    • Type: Improvement
    • Status: Won't Fix
    • Resolution: Done
    • Fix Version/s: None
    • Component/s: jointcal
    • Story Points:
      4
    • Epic Link:
    • Sprint:
      Alert Production F17 - 11, AP S18-1
    • Team:
      Alert Production

      Description

      lsst::afw::geom::SkyWcs presently has no reliable way to extract SIP terms, and that will likely be true for several more months. However, one can easily get a TAN approximation centered at any point. Is that good enough to initialize jointcal?

      Please run jointcal on a moderate size dataset with significant distortion and see if it can reliably produce a good model if the initial WCS has hasDistortion false.

        Attachments

          Issue Links

            Activity

            Hide
            Parejkoj John Parejko added a comment -

            I need to reprocess my validation_data_hsc, and run it with the constrainedModel with SipToGtransfo.cc:31 set to always false and compare. Particularly since the constrainedModel will start with an affine ccd-level model.

            Show
            Parejkoj John Parejko added a comment - I need to reprocess my validation_data_hsc , and run it with the constrainedModel with SipToGtransfo.cc:31 set to always false and compare. Particularly since the constrainedModel will start with an affine ccd-level model.
            Hide
            astier Pierre Astier added a comment -

            The first need for good WCS's is the association step. With poor approximations, it is unlikely to work properly. Dominique Boutigny has been working with poor input WCS's and may comment, but I think that we should stay below the arcsec level of approximation.
            Regarding the initialization of the constrained model, any WCS that transforms pixel coordinates into sidereal coordinates is elligible.
            The current fit initialization code does not assume a SIP distortion, but cooks up a polynomial approximation of the required degree from the input WCS. I don't think the code even assumes that the input WCS uses a TAN projection (although TAN projections are used internally). So, if WCS's good enough for associating are provided (and they do no have to be approximations), then the fit initialization does not require any extra work.

            Show
            astier Pierre Astier added a comment - The first need for good WCS's is the association step. With poor approximations, it is unlikely to work properly. Dominique Boutigny has been working with poor input WCS's and may comment, but I think that we should stay below the arcsec level of approximation. Regarding the initialization of the constrained model, any WCS that transforms pixel coordinates into sidereal coordinates is elligible. The current fit initialization code does not assume a SIP distortion, but cooks up a polynomial approximation of the required degree from the input WCS. I don't think the code even assumes that the input WCS uses a TAN projection (although TAN projections are used internally). So, if WCS's good enough for associating are provided (and they do no have to be approximations), then the fit initialization does not require any extra work.
            Hide
            Parejkoj John Parejko added a comment -

            Because Russell Owen was able to get SkyWcs to go through jointcal, we don't need to do this.

            Show
            Parejkoj John Parejko added a comment - Because Russell Owen was able to get SkyWcs to go through jointcal, we don't need to do this.

              People

              • Assignee:
                Parejkoj John Parejko
                Reporter:
                rowen Russell Owen
                Watchers:
                Dominique Boutigny, John Parejko, Pierre Astier, Russell Owen
              • Votes:
                0 Vote for this issue
                Watchers:
                4 Start watching this issue

                Dates

                • Created:
                  Updated:
                  Resolved:

                  Summary Panel