2007 August 30
I think I did a fairly good job of restraining myself and spent a minimal amount of time on overengineered plotting infrastructure today. I needed to write some code to help me see what I was doing with the HAT data, but I didn’t end up writing too much of it. I also found a SciPy implementation of 2D Gaussian fitting, which saved me some potential headaches.
I used the Gaussian fitter and some coordinate display code to get good positions for the stars of interest in the HAT data. I spent some time fiddling with that until it occurred to me that, you know, part of fitting a 2D Gaussian is deriving an amplitude, which could be kind of useful for photometry. Unfortunately, this kind of boneheaded PSF fitting wasn’t very effective. I’m not 100% sure that I was doing things correctly, but whatever I was doing yielded some really poor results. Spent some time seeing if I was just slicing the data the wrong way, but that doesn’t seem to have been the case.
I then moved from doing lame, home-brewed PSF fitting to lame, home-brewed aperture photometry. It wasn’t hard to write some quick routines to add up the pixels inside an annulus; then, doing that around the stars of interest and the sky patches, voila, instant aperture photometry. I think. Unfortunately, this also yielded sub- spectacular results. As with the PSF fitting, I wrote all the code without really checking how things are supposed to be done, which probably has something to do with the fact that my results are horrible. (I checked, and the data from the night that I’m using caught nearly a complete transit, and the lightcurve in the HAT-P-1b paper has really good precision, so I can’t just blame the data. In retrospect, I should have checked that this night was good before starting working on the reduction routines. I seem to be really good at not doing things that I know to be good ideas.)
At the end of the night, I decided to stop being dumb and found myself a copy of the Handbook of CCD Astronomy. The flimsy excuse I have for not going to it before writing code is that I couldn’t think of where to get one; but unsurprisingly there are three copies in the undergraduate lab. I’ve only started reading a few sections, but I hear that the book is good, and at the very, very least it’ll let me check that I’m actually correctly remembering how this relative photometry stuff is supposed to be done. (Or I assume it will.)
But I really need to get better at researching previous work before diving in and doing something myself. Cough, cough, plotting and numerical code. Seriously, I realize that the way I go about things right now is not smart, and yet I keep on doing it. It’s weird how blatantly irrational people can be.
(Years-later update: of course, the Greeks have a word for this phenomenon: akrasia.)