Difference Image Project (Stack Club Course Summer 2020)

[srodney]

The "difference image" interest group is considering a 3-step project:

1. data exploration with butler (how to find a dataset suitable for diff image construction)
2. making diff images (running the AP pipeline)
3. plant fake transients, and then detect them in the diff images

We are planning to use some combination of DECam and HSC data, and will build off of existing notebooks and code for building templates, injecting fake PSFs, and detecting them. (adding links below is encouraged)

[srodney]

Updated discussion of this project in the StackClub session on 2020.07.17 (with @MarkusRabus @ywx649999311 and @jbkalmbach ). Our planned project structure now looks like this

01 - Data discovery : (led by @MarkusRabus)

• given a repo with some calexps and coadds, make a table of what's there (visit number, detector, filter, obs date, etc.), write this out as a pickle (maybe in future make it something useful to the gen-3 butler).
• Make a new data directory in some scratch space defined by the user, and butler.put() the calexps there (so they can be modified in the next step)

02 - Add fakes : (led in stackclubcourse by Dani Chao)

• given a data dir with calexps, add fake stars to them
• use the ap_pipe tools to ingest the modified calexps, and make this an actual repo

03 - Make diff images : (led by @ywx649999311 )

• given a repo with calexps from 02, and a parent repo with coadds from 01, run the ap_pipe image difference driver to make diff images

04 - Visualize results (led by @srodney )

• given a repo with diff images produced by 03, showcase the results (postage stamps, light curves, etc)

[MarkusRabus]

Triggered by a comment from @kadrlica and @jcarlin. I found the following task: https://community.lsst.org/t/new-tasks-for-fake-source-insertion/3722 In this case, you can specify a new directory for the rerun. Would this task be useful for @dchau?