Uploaded image for project: 'Data Management'
  1. Data Management
  2. DM-7821

Allow for a consistent removal of peaks in the deblender when running multiBand.

    Details

    • Type: Story
    • Status: Won't Fix
    • Resolution: Done
    • Fix Version/s: None
    • Component/s: None
    • Story Points:
      8
    • Sprint:
      DRP F16-5, DRP F16-6, DRP S17-1
    • Team:
      Data Release Production

      Description

      An important part of the SDSS deblending algorithm was removing objects with similar templates. Currently, when running multiBand each filter runs the deblender independently so we do not allow it to remove objects in order to keep the same number of objects. We need to be able to consistently remove objects inside the deblender and have it be consistent across filters.

        Attachments

          Issue Links

            Activity

            Hide
            swinbank John Swinbank added a comment -

            Bob Armstrong: after some internal hand-waving, I'm marking this as 6 SPs. Could you please check and see if you agree with that estimate, and, if not, fix it?

            Show
            swinbank John Swinbank added a comment - Bob Armstrong : after some internal hand-waving, I'm marking this as 6 SPs. Could you please check and see if you agree with that estimate, and, if not, fix it?
            Hide
            vpk24 Vishal Kasliwal [X] (Inactive) added a comment -

            Looks good but please rebase pipe_tasks & obs_base off've master.

            Show
            vpk24 Vishal Kasliwal [X] (Inactive) added a comment - Looks good but please rebase pipe_tasks & obs_base off've master.
            Hide
            swinbank John Swinbank added a comment -

            Bob Armstrong — thanks for providing nice task documentation here!

            Show
            swinbank John Swinbank added a comment - Bob Armstrong — thanks for providing nice task documentation here!
            Hide
            swinbank John Swinbank added a comment -

            Hey Bob Armstrong: looks like this was reviewed quite a while ago. Anything delaying the merge?

            Show
            swinbank John Swinbank added a comment - Hey Bob Armstrong : looks like this was reviewed quite a while ago. Anything delaying the merge?
            Hide
            swinbank John Swinbank added a comment -

            Following discussions between Bob Armstrong, Jim Bosch & others on 2016-12-02, and again at the Princeton group meeting on 2016-12-05, we have decided to change the approach being taken to multi-band deblending. The new technique will involve serializing subsets of the footprints in each patch to separate output files, and then deblending them separately. The subsets will be chosen such that each subset can be processed simultaneously across all bands (ie, without running out of memory, except with occasional & unavoidable very large footprints). This work will pave the way for a "true" multi-band deblender, and, as such, will take place on DM-8140.

            Show
            swinbank John Swinbank added a comment - Following discussions between Bob Armstrong , Jim Bosch & others on 2016-12-02, and again at the Princeton group meeting on 2016-12-05, we have decided to change the approach being taken to multi-band deblending. The new technique will involve serializing subsets of the footprints in each patch to separate output files, and then deblending them separately. The subsets will be chosen such that each subset can be processed simultaneously across all bands (ie, without running out of memory, except with occasional & unavoidable very large footprints). This work will pave the way for a "true" multi-band deblender, and, as such, will take place on DM-8140 .

              People

              • Assignee:
                rearmstr Bob Armstrong
                Reporter:
                rearmstr Bob Armstrong
                Reviewers:
                Vishal Kasliwal [X] (Inactive)
                Watchers:
                Bob Armstrong, John Swinbank, Vishal Kasliwal [X] (Inactive)
              • Votes:
                0 Vote for this issue
                Watchers:
                3 Start watching this issue

                Dates

                • Created:
                  Updated:
                  Resolved:

                  Summary Panel