As expected (by those wiser than me), most of the sources that are detected in GALEX but not SDSS are small technical problems with either catalog. In many cases they are GALEX artifacts. In others they are SDSS dropped fields or dropped sources (usually because they are too close to the edge of the survey boundary). A lot are GALEX sources in the faces of big, nearby galaxies, which GALEX software splits up into sub-sources differently than SDSS (see an example below, in an enormous nearby galaxy—can you tell which one?). So I wrote code today to catch all of these cases. It is running now. I hope we find some true GALEX-only sources.
In demonstrating a purely technical technique for measuring proper motions in multi-epoch data, I may have accidentally discovered about 100 L and T dwarfs in the local neighborhood today. Jester and Rix are helping me to sort it out. If I have, it will de-rail my technical progress because I will have to stop and write a scientific publication.
Among other small tasks for a number of ongoing projects, I made figures of some GALEX sources that are inside the SDSS area but not detected in SDSS. Here's an example, with the GALEX position circled with a 3-arcsec radius circle (roughly the error circle). Then, once I made these images, I realized I hadn't employed any of the standard cuts you use to remove spurious GALEX sources from the automatically generated GALEX catalogs! D'oh. So statistics have to wait for tomorrow. Schiminovich: cuts, please.
I finally got around to plotting the proper motions of 6000 extremely red stars in the SDSS Southern Stripe today. There are many with very large proper motions, going nearly to arcsec per year levels! If these are not errors (and they easily could be), they are likely to be cool dwarfs, either very nearby or else in the halo. Either way, they are interesting. There are also many with very small proper motions; these are candidates to be high-redshift galaxies and quasars.
I spent part of the day carefully reading a paper by Wu on her analysis of the mid-IR and spectral properties of some extremely low luminosity galaxies in the SDSS. She finds that the deficit of PAH emission in low-luminosity star-forming galaxies is related to the metallicity and to the radiation fields in the galaxies, but both of these properties are closely related to one another and to many other properties, so the causal relations are not yet known, in my view.
Fought household illnesses all of the long weekend and today, but nonetheless managed to make a huge plot of everything vs. everything in the GALEX match to SDSS. That plot clearly shows the stellar locus, hot white dwarfs (I think), and AGN. Today I worked on understanding the clever NYU-VAGC description of the SDSS survey geometry so as to figure out which GALEX sources are inside the SDSS region. Once this (incredibly slow) process is complete, we will be able to discover sources that are bright in GALEX but not visible at all in SDSS. Should any such things exist? If there are any true non-detections in SDSS of true GALEX detections, they would be interesting, either as transients or sources with wacky spectra.
My research today—shepherding along the huge GALEX–SDSS match—was so boring that I can't describe it here. Given the usual standards I maintain at this blog, that's saying something! What I can say is that at journal club I learned about a very tantalizing Solar-System-like exoplanetary system discovered in part with observations by amateur astronomers. That's a great application for the Open-Source Sky Survey! (A challenge: try to guess our URL.)
In thinking about creating a generative model of every astronomical image ever taken, the Astrometry.net team has been considering what is possible at
catalog level; that is, what can you do with the catalogs alone, without going to pixels. In the spirit of exploring that, Schiminovich and I started by just matching the GALEX catalogs to the SDSS catalogs. This will not produce the most sensitive possible match because there may be sources in each catalog that are below the detection limit but nonetheless detectable given prior information from the other catalog. I got to the point of an angular distance distribution for GALEX vs SDSS; as Schiminovich predicted, it has a width of about 2 arcsec.
In the first baby step towards a multi-wavelength all-sky catalog of point (compact) sources, Schiminovich and I decided to look at the point sources that are bright in the GALEX FUV band. The first question we will ask is:
Is it possible to predict with good confidence that a star will be bright in GALEX FUV given only the SDSS (and maybe 2MASS) fluxes? We began by writing scripts to read a subset of columns from the GALEX and SDSS pipeline outputs, so that we can concatenate all point sources from each survey into individual, monstrous files.
At lunch, Young Wook Lee (Yonsei) gave a nice talk on the blue horizontal branch in Milky Way globular clusters, showing that its morphology and population is a very strong function of metallicity and, in particular, helium abundance. It appears that many of the Milky Way's most dense globular clusters have multiple populations with different helium abundances. This is hard to explain, but Lee hypothesizes that these may not be what we traditionally think of as globular clusters but rather the remnants of disrupted substructures accreted onto the Milky Way long ago.
Zolotov and I spent the afternoon planning her first paper and her dissertation, in preparation for her oral exam. Her first paper will focus on the kinematics of stars with different origins (formed in situ, accreted recently, accreted long ago) in a simulated Milky-Way-like galaxy.
I spent some enjoyable time working through Bovy's write-up of his work on the transparency or dust content of galaxy clusters (he is comparing mean spectra of red galaxies
not behind clusters taken from Berlind's group catalog in SDSS. As expected, we have very sensitive limits on the dust content of the clusters, and it also appears that we have a positive result. But the big challenge is to compute realistic error bars; most bootstrap or jackknife methods will underestimate them.
This weekend I returned to my summer project of distinguishing faint quasars and brown dwarfs via proper motions. I made image-stacking code that compares the stack at zero proper motion with the stack at best-fit proper motion. Unfortunately, the images are so low in signal-to-noise, there is very little difference between the stacks. But I now have tons of data analyzed so it is time to write it up.
I worked out derivatives and second derivatives of chi-squared as a function of date to analyze some of Barron's results in the
blind date paper. Basically, as you might expect, the information in a single star increases with the magnitude of its proper motion and decreases with the uncertainty with which it has been measured. Duh! There are some subtleties in the limit that all the stars in the field are moving in the same direction, or in the limit that there are very few stars, or for stars which are large outliers in position. Barron made some plots to check my analysis and we may be close to converging.
Wu and I discussed her thesis plan, in preparation for her qualifying exam. We tentatively settled on something built around star formation in galaxies and the processes that terminate it. But this is the beginning of a long process.
The entire family being sick, plus babysitter too, is not conducive to research getting done! Nonetheless, Barron and I got in a good hour or so on the blind date paper—the paper on determination of the date at which an image was taken, based solely on the motions of the stars. There is a big difference between his analytical calculation for the contribution of each star to our information about the image date and his brute-force numerical calculation. This difference has to be explained if we are going to publish. I promised that I would do some related math when I can get a breather.
Much as I like to rag on SQL databases, they are indispensible in dynamic web applications (recall how
web 2.0 we are here at Astrometry.net), I am happy to report that Lang and I spent most of the day discussing the database schema for the web version of our automated calibration. The challenge is to stay close to what's known in the bidness as