Uploaded image for project: 'Request For Comments'
  1. Request For Comments
  2. RFC-526

Add FGCM as a 3rd party package to the stack and move fgcmcal from lsst-dm to lsst

    XMLWordPrintable

    Details

    • Type: RFC
    • Status: Implemented
    • Resolution: Done
    • Component/s: DM
    • Labels:
      None

      Description

      The Forward Global Calibration Method (FGCM) performs global photometric survey calibration with forward modeling of the atmosphere and instrument, including chromatic corrections (see http://adsabs.harvard.edu/abs/2018AJ....155...41B for the paper describing the method).

      When DM-14670 is complete, the stack will be able to run the full FGCM calibration on (at least) HSC data, with outputs fully compatible with stack products produced by (e.g.) jointcal. FGCM has been successfully run on HSC PDR1 data (DM-14669) and is also being run on HSC S18a data (DM-15545). The only limitation for running on other cameras is the requirement that transmission files be available (DM-12366).

      In order to bring this fully into the stack, the FGCM (pure python) third-party package (https://github.com/erykoff/fgcm) will have to be added to the stack (via a github fork to allow easy continued development), and the fgcmcal package currently hosted by lsst-dm (https://github.com/lsst-dm/fgcmcal) will have to be moved to lsst_distrib. There are no additional dependencies that will have to be added that are not already in lsst_distrib.

        Attachments

          Issue Links

            Activity

            Hide
            tjenness Tim Jenness added a comment -

            Ok with me.

            Show
            tjenness Tim Jenness added a comment - Ok with me.
            Hide
            Parejkoj John Parejko added a comment -

            +1

            Is the plan to treat lsst's fgcm as an actual fork, or to always port changes between the forks? Which fork will be the one where development primarily occurs?

            Show
            Parejkoj John Parejko added a comment - +1 Is the plan to treat lsst's fgcm as an actual fork, or to always port changes between the forks? Which fork will be the one where development primarily occurs?
            Hide
            erykoff Eli Rykoff added a comment -

            I believe the plan is that the development occurs outside the fork in the third-party package, and we just use the stack fork to allow for quicker/easier updates than the old tarball method which was pretty cumbersome. The reason for the local fork is that changes upstream to master won't affect the stack version without an explicit choice to make the synchronization.

            As it is, I have an eups table in the third-party fgcm package so it can be setup normally (with a -r path), so this part should be easy. It's just not currently distributed with the stack.

            Show
            erykoff Eli Rykoff added a comment - I believe the plan is that the development occurs outside the fork in the third-party package, and we just use the stack fork to allow for quicker/easier updates than the old tarball method which was pretty cumbersome. The reason for the local fork is that changes upstream to master won't affect the stack version without an explicit choice to make the synchronization. As it is, I have an eups table in the third-party fgcm package so it can be setup normally (with a -r path), so this part should be easy. It's just not currently distributed with the stack.
            Hide
            tjenness Tim Jenness added a comment -

            Eli Rykoff it sounds like you can adopt this RFC. Please create a triggered DM ticket for implementation.

            Show
            tjenness Tim Jenness added a comment - Eli Rykoff it sounds like you can adopt this RFC. Please create a triggered DM ticket for implementation.
            Hide
            erykoff Eli Rykoff added a comment -

            RFC has been adopted, to be implemented on DM-16054

            Show
            erykoff Eli Rykoff added a comment - RFC has been adopted, to be implemented on DM-16054

              People

              Assignee:
              erykoff Eli Rykoff
              Reporter:
              erykoff Eli Rykoff
              Watchers:
              Eli Rykoff, Ian Sullivan, John Parejko, John Swinbank, Tim Jenness
              Votes:
              0 Vote for this issue
              Watchers:
              5 Start watching this issue

                Dates

                Created:
                Updated:
                Resolved:
                Planned End:

                  Jenkins

                  No builds found.