Today we inspected, in the context of the deepest imaging we have, the sources we
harvested from the Flickr group images yesterday. They look real, so we are emboldened. We realized that between this and the stuff Lang did on Comet Holmes (he can determine the orbit from the footprint of Flickr-submitted images!), we have some good evidence that Flickr and amateur astrophotographers can have a big impact on science.
Today we inspected, in the context of the deepest imaging we have, the sources we
Stumm, Lang, and I gathered some (heuristic) evidence for sources in the region of M42 (the Orion Nebula) using images submitted to the Flickr pool. The image below shows in transparent blue symbols the detected sources in the Flickr images in a tiny patch in the nebula. We take it from this that the amateur data do have highly redundant (and therefore reliable) information about many sources.
I had a great time at the thesis defense of Marilena Loverde (Columbia) today. Her thesis was on magnification bias effects on precise cosmology observables, and also signatures of non-gaussianity. She took a theory-centric view, which permitted her to investigate many precise effects, all of which will be either confusing or useful to precise surveys in the near future.
Christopher Stumm (Microsoft) and Lang are in town this week, to see if we can do reliable science with unreliable data. Stumm designed, wrote, and operates our Flickr pool in which we calibrate amateur astrophotography just as if it were professional data. Now we want to use those data as if they were professional data, since they certainly contain tons of information. But the challenge is accounting for the fact that data of unknown provenance will, in general, contain unknown artifacts, errors, and noise. We started on project specification this morning.
The plot below (made by Schiminovich and me today) shows the GALEX–SDSS colors of quasars with photometric redshifts between 2 and 3.5. The white points have spectroscopic redshifts in that same range; the cyan points don't. The vertical axis is a GALEX–SDSS color that separates them well. This might have implications for SDSS-3 quasar target selection, which takes place in regions of the sky with good GALEX coverage.
I got all fired up about decision theory this past weekend, and wrote something about it today, in the context of the Astrometry.net
verify step. Here is the first paragraph:
All science is really decision theory, not inference! After all, the end result of an inference is a paper, and what one writes in the paper depends on the result of the inference, and not at all trivially. That is, the content of the paper is not simply a statement of the posterior distribution function (obviously). Despite the obviousness of this, most discussions of scientific data analysis end at (or even before) the inference step, and don't give quantitative guidelines for anything that follows, except perhaps for the construction of confidence intervals.
Many assume that I am a religious or dogmatic Bayesian, because I use Bayesian statistics. I am not; I use them because (a) they are not ad-hoc, (b) they are justified in the context of probability theory, and (c) for many problems I do they perform better than the alternatives (see, for example, our April Fools' paper). Bovy and I spent some time today doing battle with Kyle Cranmer (NYU) on this point, which was productive inasmuch as we understood the differences between a Bayesian and frequentist approach, but unproductive in that Cranmer thinks he could do much better with a frequentist method (to our Fools' problem). We asked him to
put up or shut up, as Gruzinov says. While we were arguing, who should walk in but Yuri Levin (Leiden), the author of the frequentist orbital roulette paper.
Barry Madore (OCIW) gave a very provocative talk about the possibility that there might be dark galaxies and then showed that after some pretty hard looking (using galaxy–galaxy interaction signatures), he can't find any. So they are rare, at the near-L-star level. Nice and fundamental!
Rix, Bovy, and I discussed re-doing the famous Oort problem on the vertical structure of the Milky Way disk; that is, infer the density of the disk by simultaneously fitting for the position and velocity distribution of stars in the vertical direction. The differences now relative to the 1930s is that we have far more stars, far better measurements, and we can do the test over a finite fraction of the Milky Way disk. We might be able to see gradients and look for signatures of departure from axisymmetry, which would be qualitatively new.
Masers are incredibly bright, compact emission regions in the Galaxy that permit measurements of full six-dimensional phase-space position. Reid (Harvard) has a sample of 18 that he used to measure the mass of the Milky Way. Rix, Bovy, and I spent an inordinate amount of time criticizing the Reid method, although we are not sure we can do better. We also involved Johnston and Weinberg (UMASS) in our discussions.
Koposov may have evidence that the velocity of the Grillmair stream is not exactly parallel to the configuration direction of the stream. Since the stream is cold and very near pericenter (by any reasonable model), this could have important implications for either the potential of the Milky Way or else how the stream formed. Unfortunately, it is hard to tell which. Had many confusing conversations about this today, but Bovy worked out that there are large classes of potentials in which the misalignment can't happen unless there is either a strong action gradient in the stream or else a time-dependence (or other nastiness) in the potential.
Bovy finished another draft round on our first Hipparcos papers, one an immense method paper and one with some results, using Geneva-Copenhagen Survey to externally vet any Hipparcos-inferred structure in the velocity distribution. We find that this conservative method disfavors some of the less-significant structures claimed in the literature, and that there are no moving groups strongly evident in the Hipparcos data that weren't known prior to Hipparcos. Bad news for Hipparcos, but it does mean that we have something useful to contribute to this literature.
In the astro seminar today, Rix gave disturbing evidence that the disk spiral patterns have a fixed (or close-to-fixed) pattern speed. This was not his primary goal, but it was the most remarkable part of it all, from my point of view. One clever thing about the research is that they were able to make this argument without definitively identifying the spiral pattern or spiral arms themselves.
Bovy, Rix, and I further specified the stream-direction question, and gave it to Bovy as homework. He completed his homework by late in the evening. In potentials with high symmetry, the stream direction will be in the orbital direction. But it remains to see if it will in, say, an oblate logarithmic potential (or worse).
Rix arrived for two weeks while I was out on a mental-health break (hence no posts for a few days). Today, he, Bovy, and I spent a good chunk of the morning going over the Rix-related on-going projects, including the Solar System dynamics, Galactic Black Hole, Galactic masers, and cold stellar stream projects. In regards the latter, Rix encouraged me to write up my concerns about the direction of the tidal stream in a short note. I started that late in the day. I will say more soon, but my concern is not about the fact that there is an action gradient along the stream, my concern is that even in the limit of zero action gradient, the stream will not obviously (to me, anyway) stretch out in the orbital direction. That said, it usually does in the simulations!
After a frustrating time looking for a units bug (there wasn't one) in Wu's code, I went to Zeljko Ivezic's (UW) talk at the Columbia Astronomy Colloqium. He spoke about LSST and the relationship to SDSS, intellectually, technically, and scientifically. It was great (though I had to leave a few minutes early).
Prior to that, Schiminovich and I got close to a measurement of the ionizing radiation incident on the outskirts of the Milky Way from optically selected quasars. It looks from the data (and from models too) like we can straightforwardly extrapolate to 912 A (observed frame) despite the intergalactic medium.