| | 8 | This week was short with several people taking an extra day or two holiday along with Thanksgiving. We (mostly Heather) got some processing done, more-or-less keeping up with the nightly science data (what there was), and nearly finishing off re-processing of the MD data. I have been digging into ppSub + psphot interaction issues. I started to work with a set of images from the Run 4 3pi survey data, but decided to narrow my scope a bit to make faster progress. I generated a pair of images for the simtest camera with the same pointing and stars, but with a small number of additional stars in one image (fake transients). The small size allows me to run through the ppSub + psphot steps quickly including visualization. I wanted to check if the choice of Poisson or constant errors had a significant impact on the photometric consistency, but the choice of weighting was not the smoking gun I had hoped; I am still investigating the impact of PSF-mismatch on the photometry. I am still not certain what are the sources of error in the difference photometry. I did identify a couple of second-level bugs (important, but not the main concern), and some other issues that I will continue to explore in the coming week. In particular, the dual convolution analysis seems to result in a more-deviant output PSF than the single convolution, for very similar input PSFs. |
