(Un)recognized blend metrics in BTK for BLxPZ and BLxCL

[xuod]

(Un)recognized blend metrics in BTK for BLxPZ and BLxCL

Implement metrics to quantify detection mismatches due to recognized and uncrecognized blends in BTK and connect to BLxCL and BLxPZ activities.

Contacts: @xuod @ismael-mendoza @aimalz Day/Time: Main communication channel: #desc-blxpz #desc-blxcl GitHub repo: https://github.com/LSSTDESC/BlendingToolKit and https://github.com/LSSTDESC/lost-and-found Zoom room (if applicable):

Goals and deliverable

• Definition of relevant metrics
• Implementation in lost-and-found and interface with BTK and RAIL

Resources and skills needed

• BTK developers/enthusiasts
• RAIL developers/enthusiasts
• other BL/PZ/CL/WL enthusiasts

Detailed description

[add additional details about the project]

[aimalz]

@ismael-mendoza and I are preparing by making some mock data sets to play with next week, specifically the catalogs resulting from a couple deblenders (SourceExtractor threshholding with two different threshold values) and a couple cross-matchers (kdtree with two different distance values) to make the metrics more concrete, so we don't end up hunting down edge cases that really don't come up in plausibly realistic data. We'll be sprinting in the afternoons Wednesday-Friday.

[ismael-mendoza]

In case anyone is curious, here is the notebook showing how mock data is generated: https://github.com/LSSTDESC/BlendingToolKit/blob/im/bkt-mock-data/notebooks/mock-data.ipynb

[EiffL]

A little piece of random info. There is a fork/extension of galsim, which does flexion: http://physics.drexel.edu/~jbird/index.html

It would be fantastic I think to add these features into the upstream galsim

[sidneymau]

Related to Francois' comment – https://github.com/GalSim-developers/GalSim/issues/361