order-of-magnitude statistics

I figured out a relatively principled way for us to decide, when we try to solve an astronomical image blind, whether or not our solution is the correct one. We want to be conservative (false positives lose us credibility) but we don't want to throw out the baby with the bath water. I figured out a criterion that will keep our false positive rate at the 10⁻⁶ level—to order-of-magnitude. Exact statistics are not possible here because the stars on the sky are not uniform (not a Poisson process), and the moments of the distribution are not known (they could be measured for the standards catalog, but they can't be known for arbitrary input images).

In other news, Lang got the blind system to solve this image today. Outrageous!

Sagittarius stream

Late today it became apparent that Zolotov has a measurement of the velocity of the Sagittarius stream from the spectroscopically observed stars in the SDSS. The SDSS mainly took spectra of galaxies, but it had some categories for interesting stars, including blue horizontal branch and blue straggler stars in the Milky Way halo.

paper 1

I worked on astrometry.net's paper 1 (the paper for the astronomical community) while Mierle worked on streamlining our object-detection code.

kinematic and spatial MW substructure

Zolotov and I talked about creating a micro-project on kinematic substructure in the Milky Way halo. Lots of substructure has been found in real space, and lots of stellar velocities have been measured, so we are looking in velocity space. But there are many data issues—most kinematic samples are horrendously selected. We want to start with a baby step that is insensitive to these selection effects. I think we will start by measuring kinematics for known structures.

going alpha

We are getting close to the astrometry.net alpha release. I spent my research time today finishing the tests I described yesterday; our internal code passed with flying colors (it actually centroids the stars more reliably as a function of signal-to-noise than I expected). Lang has been doing a power of work duct-taping code together, getting it all to talk to the web, and making indices for images of all scales. This week, in his spare time, he solved blind two of the astronomy pictures of the day, one of the vela nebula and one of the rosette nebula (below)!

Yes, he just uploaded that JPEG into our astrometry.net pre-alpha web page and astrometry.net gave him back its location and orientation on the sky.

fake data for astrometry

I spent a little time working on making fake data for functional testing of the internals of astrometry.net. In particular, we started to get afraid that we aren't completely consistent with the FITS convention that the center of the first pixel in an image array get (x,y) coordinates (1.0,1.0). We have software kicking around that makes the center of this pixel (0.5,0.5) and most of our IDL code makes it (0.0,0.0), but the FITS convention is what it is, and we love and obey standards. The best way to ensure compliance with the standard is to produce a fake image with known star positions and run the code on it.

scalar objective

My only research contribution today (besides attending some nice talks on particle experiments) was to start the discussion on astrometry.net about what scalar objective we ought to optimize in the tweak stage. It is not completely trivial because the usual kind of objective (things like chi-squared) can be optimized simply by rejecting pesky outliers. These days, my position has been that you haven't made a measurement of anything unless you have optimized some specified—and justified—single-valued figure of merit.

meta-development for astrometry

All research time this weekend was spent with Mierle, working through all of the outstanding tickets on the astrometry.net bug tracking system. We are trying to make the tickets atomic, specific, prioritized, and correctly categorized by short-term milestone. We also started the process of creating use cases or customer stories to guide our decision making by focusing on what we need to provide to our anticipated user base (rather than just what is most fun!). At the end of the weekend we spent some time brainstorming about the far-future capabilities of astrometry.net—we couldn't type fast enough to keep up with the good ideas we have.

OLPC

This isn't 100 percent research, but today Sanjoy Mahajan (MIT) brought by an OLPC and we tested out my nascent OLPC Planetarium software. That was fun!

many small things

I looked at the line emission from massive galaxies in the SDSS for a few minutes this morning, in preparation for understanding their evolution in the three-dimensional space of luminosity, color, and central stellar density. My main question is whether and how I can correct the H-alpha line equivalent widths for contributions from LINER or AGN emission to get a purer measure of the star-formation rates.

Keir Mierle (Toronto) and Dennis Zaritsky (Arizona and NYU) split group meeting today, with Mierle giving a status report on astrometry.net (which is rocking) and Zaritsky talking about the dynamical or kinematic continuum connecting elliptical galaxies to central cluster galaxies, dwarf spheroidals, and spiral galaxies (which is remarkable).

Frank van den Bosch (MPIA) gave our astro seminar on the occupation of dark matter halos by galaxies, and the predictions and power of models that describe the Universe in this way. After his talk I gave him a copy of my latest paper on galaxy environments (for bed-time reading), because he made the very nice point (with which I agree) that we should be using halo or group mass as the environment measure from now on. We will continue that discussion this summer when I am at MPIA.

Late in the day, Blanton and I went on a PRIMUS phonecon, in which we discussed the current status, which now includes some very promising redshift determinations, thanks to work by Burles, Bolton, Blanton, and Cool. We still don't have redshifts for the majority of our 300,000 spectra, but we think most of the issues come down to our throughput model, which is evolving daily.

Virgo overdensity

In our quest to understand the possibility of measuring the one-dimensional and three-dimensional kinematics of Local Group substructure with stellar spectra, Zolotov and I discussed the enigmatic Virgo overdensity today. The debate over its origin could be ended with a good velocity. I feel certain someone has already measured it, but it is a good exercise to measure it with the data we have in hand nonetheless.

Milky Way and normal galaxies

I had a great trip to Columbia yesterday. I spent most of the time up there discussing halo substructure with Eric Bell (MPIA), Kathryn Johnston, Sanjib Sharma, and James Bullock (UCI; by phone). Bell has a nice paper with a variance (power spectrum) analysis of substructure in the Milky Way halo, finding that there is a lot of variation around any smooth fit. Johnston and Bullock find the same in simulations. Before and after this I discussed with Zolotov our project to look at substructure in the velocity distribution. Today, Willman (now CfA) suggested we check out the super-awesome RAVE public samples, which contain 10⁵ stars with well-measured radial velocities.

While I was up there, Schiminovich got me fantasizing again about a huge sample of normal galaxies with Spitzer spectra. The Spitzer deadline is in just a few days. But the point would be to make up the difference between all the specialized samples out there (star-forming galaxies, Seyferts, quasars, dwarf galaxies, etc.) and a complete, flux-limited sample.

H-alpha measurements

Yesterday was a low-research day, but I did re-start Quintero's ancient code to measure H-alpha line measurements and A-star and K-star contributions to galaxy spectra on all of the SDSS spectra to date. Despite the age of the code and subsequent changes to the SDSS data reduction pipelines and outputs and the fact that it has to run without crashing on 1.2 million SDSS spectra, Quintero's code just works. Now that's what I like about my collaborators! I need the spectral measurements to compare with the central stellar densities in my project to take a stab at a "continuity analyis" of the galaxy population.

Today I am off to Columbia to work with Bell and Johnston on our nascent paper on substructure in the MW halo.

galaxies and precision cosmology

In a very special group meeting, Anja von der Linden (MPA) and Eric Bell (MPIA) gave back-to-back short talks on galaxy evolution. von der Linden is looking for evidence that brightest cluster galaxies have a different formation mechanism from field or ordinary galaxies of the same mass and color. Her control sample is so nicely composed, she can rule out the null hypothesis at good confidence, despite the fact that her primary result is that in fact the BCGs are amazingly similar to galaxies of the same mass and color in other environments.

Bell asks whether the evolution in the total mass density in red an blue galaxies can be explained with passive models, or whether merging is necessary and sufficient. He finds both, ie, that merging is both necessary and sufficient to explain the evolution of the red sequence, with the caveat that everything has large error bars.

In the afternoon, Lam Hui (Columbia) showed us various issues with doing precision (and accurate) cosmology with the next generation of huge surveys, including baryon acoustic feature, which is close to our hearts here at NYU. He gave us some great stuff to look for with PRIMUS.

oblateness vs prolateness

I realized today (or yesterday now I guess) that as a side project to my project on the evolution of galaxies that might evolve into luminous red galaxies, I can, with very little additional work, produce the necessary output to look at the question of whether the LRGs are oblate or prolate, as suggested by Scott Tremaine (IAS). Tremaine's idea is clever because it does not require accurate deconvolution or modeling of the axis ratio distribution; it uses the fact that the galaxies are transparent and asks about relationships between apparent axis ratio and projected stellar density. I did the necessary work to make sure the data that need to be carried along with the evolution data are indeed carried along.