adrn / achtungcoaddverboten

Hey 👋, I'm Adrian (@adrn)

I am an associate research scientist (ARS) at the Flatiron Institute where I help lead the Nearby Universe group and study the structure and dynamics of stars and dark matter in our galaxy (the Milky Way 🌌). I am also the Assistant Director of Scientific Software at the Simons Foundation, and am on the Astropy Coordination Committee. I do much more 👨🏽‍💻 than 🔭!

• 🕸️ Personal website
• 👥 CCA Nearby Universe Group
• 🟦 Blue Sky
• 🧻 Google Scholar
• 👨🏽‍🔬 CV
• 👨🏽‍💼 Resume

achtungcoaddverboten's Issues

Write the abstract

Do it!

emphasize ease

DFM points out that the key point to make is not just that coadding is bad, but that not coadding is nonetheless easy to do / live with.

Build infrastructure for multiple trials

Right now I have no way to save data for multiple trials. I should make some large NTrials x NImages array to store information for each of our sorting / weighting experiments.

Write a short abstract for the README

Write a short abstract for the README...

Make paper outline / plot attribute table

Make some version of the attached image in the latex doc.

Smoothing images screws with sorting the smoothed images

Smoothing the images sets all image.star.sigma equal to the same value, so they can't be sorted for PSF ordering. This messes with my sorting so I'll have to sort on the PSF of the parallel list of unsmoothed images somehow.

Add a 'wiggle' flag

Add a flag in the code to allow the true centroid of the star to wiggle around a little bit between coadds.

Add a difference between showing and saving plots

Right now the --plot flag shows some plots, saves (silently) some others...make it more obvious what the code is doing!

Star position should vary

The "star" position should be different between trials, but the same for a single stack of images. Currently, it is always placed in the exact center of the image, but we should move it around a little bit. It should always stay within a few pixels of the center of the image however because we will typically use a small image size (16 pixels by 16 pixels).

Add flag to modulate sources flux

We should do 3 cases:

• ~3.5*sigma in PANSTARRS
• ~6*sigma in PANSTARRS
• ~3.5*sigma in LSST

Computing positional chi-squared is wrong

For some reason, the measured position is always very close to the center of the pixel that the star is placed in. I think this is a problem with computing the chi-squared grid.

Outline the paper

Add an outline to each section in the paper.

change name -- too many hyphens

how about StopCoadding or DoNotCoadd or AchtungCoaddVerboten