Dustin Lang

My research is at the intersection of computer science and astronomy. I mostly work on principled data analysis methods for astronomical images, drawing on old and new results from computer vision, machine learning, signal processing, information theory, and statistics. On the astronomy side, the goal of my work is to make the best possible use of the data we have to answer scientific questions and to discover new questions to be asked. On the computer science side, I use astronomical images, with their beauty and underlying simplicity, as a motivator for the development of new machine learning and inference models and methods.

Lately, I have been working on combining imaging datasets taken by different astronomical instruments in order to benefit from their complementary wavelength coverage, sensitivity, and resolution. Specifically, I have been focusing on simultaneous analysis of imaging data from the Wide-Field Infrared Survey Explorer (WISE) satellite and ground-based optical imaging from the Sloan Digital Sky Survey (SDSS). SDSS provides nearly ten times better resolution, which allows us to interpret the WISE imaging by resolving nearby stars and galaxies that are blended together in the WISE images. For this work, we have developed a generative model for astronomical images, dubbed the Tractor. As part of my work with the WISE data, I had to produce a new set of combined images (coadds, in the astronomical jargon) from the multi-exposure WISE data; this project is called unWISE.

My results are being used to select quasar and Luminous Red Galaxy targets for spectroscopy in the SDSS-3/SEQUELS program, and the upcoming SDSS-4/eBOSS program, which will map huge swaths of the high-redshift universe to make measurements of the baryon acoustic oscillation scale and nail down the growth history of the universe. The generative image modelling work we are doing with the Tractor will be used for imaging data analysis and targeting of the Dark Energy Spectroscopic Instrument (DESI) project; we are actively collaborating with statisticians from Berkeley and Harvard to explore the large-scale inference challenges presented.

I support and have research interests in amateur astronomy, through the Astrometry.net project (in particular the web service nova.astrometry.net), and through some projects exploring how we could use the data produced by amateur astronomers for scientific purposes, including reconstructing the orbit of Comet Holmes using web image search and combining images of unknown provenance and calibration into an open-source sky survey. [more]

DESI Imaging Legacy Surveys: image analysis
We are conducting a public imaging survey of over 15,000 square degrees using the Dark Energy Camera (DECam) on the Blanco telescope in Chile, the Mosaic3 camera (now decommissioned) on the Mayall telescope in Arizona, and the 90prime camera on the Bok telescope in Arizona, in g, r, and z bands, to about 2 magnitudes deeper than SDSS. Together, these surveys are called the Legacy Surveys. They provide a valuable legacy survey for a wide range of science projects. I am leading the efforts to reduce these data using the legacypipe code, a framework built around the Tractor, a code for generative image modelling of multi-band, multi-epoch data sets.

Browse these spectacular images using our Imagine Sky Viewer.

unWISE coadds
I released a reprocessing of the WISE infrared satellite images at the full available resolution. We need this for doing forced photometry in the SDSS and Legacy Surveys contexts, but others may find them useful. Later, Aaron Meisner continued this effort, and has produced full data releases each year as WISE has continued to deliver imaging in W1 and W2 bands. Details: unwise.me.

CHIME/Fast Radio Bursts
I am part of the CHIME/FRB collaboration, who are using the Canadian Hydrogen Intensity Mapping Experiment (CHIME) radio telescope, located not far from my home town in British Columbia, Canada, to detect Fast Radio Bursts (FRBs). FRBs are mysterious, brief bursts of radio emission that have the signature of coming from vast distances, and therefore appear to be very energetic. With CHIME/FRB, we are collecting unprecedentedly large samples of FRBs to help unravel this mystery.

Older Projects

SDSS-III/SEQUELS targeting via forward modelling
The SEQUELS program will targets 800 square degrees of sky, taking spectra of Luminous Red Galaxies, Emission Line Galaxies, and quasars. The LRG and quasar targets are distinctive in the mid-infrared as observed by WISE. We are using our generative (forward) modelling approach, the Tractor, to do a simple type of joint analysis of the ground-based optical SDSS imaging data and the space-based infrared WISE data. In this "forced photometry" approach, we take the positions and profiles of all SDSS objects as fixed, compute a pixel-level model for what these objects would look like in the WISE images, and then find a weighted combination that best fits, in a least-squares sense, the observed WISE images. This is a generative model of the WISE images in which many of the parameters are held fixed.

Panchromatic Hubble Andromeda Treasury (PHAT) survey
I handle the astrometry for PHAT, a multi-year survey of the Andromeda galaxy using three cameras aboard the Hubble Space Telescope. I am responsible for the precise alignment of the over 12,000 exposures of the survey, which allows us to build seamless mosaics and catalogs of hundreds of millions of stars to be used by the survey science teams.

Large Synoptic Survey Telescope (LSST)
I wrote and maintain the deblender for the LSST stack. A deblender tries to do source separation of astronomical objects in images: disentagling overlapping stars and galaxies so that the individual objects can be measured separately.

LSST is also using our Astrometry.net code for initial astrometry.

Astrometry.net
I continue making small improvements to the Astrometry.net code and the Nova.astrometry.net web service. Astrometry.net is a system for automatically recognizing (or astrometically calibrating) astronomical images using pattern recognition and fast search techniques. We have thousands of users who have uploaded tens of thousands of astronomical images.

This web page is often stale, but my arXiv listing usually contains all my astronomy papers.

See my CV for the full list.

In Prep / Under review
Published (or nearly so)
Lang, D., Hogg, D.W., and Schlegel, D.J., 2014, "Photometry of 400 million SDSS objects in the WISE imaging," Astronomical Journal

Lang, D., 2014, "unWISE: unblurred coadds of the WISE imaging," Astronomical Journal, 147, 108. arxiv:1405.0308

Lang, D., Hogg, D.W., and Schölkopf, B., 2014, "Towards Building a Crowd-Source Sky Map," AI & Statistics 2014, JMLR Workshop and Conference Proceedings, 33, 549. arxiv:1406.1528

Hogg, D.W. and Lang, D., 2012, "Replacing standard galaxy profiles with mixtures of Gaussians," Publications of the Astronomical Society of the Pacific 125, 719. arxiv:1210.6563

Foreman-Mackey, D., Hogg, D.W., Lang, D., and Goodman, J., 2012, "emcee: The MCMC Hammer," Publications of the Astronomical Society of the Pacific, 125, 306. arxiv:1210.6563

Bundy, K. et al., 2012, "SynMag Photometry: A Fast Tool for Catalog-Level Matched Colors of Extended Sources," The Astronomical Journal 144, 188. arxiv:1301.3164

Weisz, D.R. et al., 2012, "The Panchromatic Hubble Andromeda Treasury IV: A Probabilistic Approach to Inferring the High Mass Stellar Initial Mass Function and Other Power-law Functions," The Astrophysical Journal, 762, 123. arxiv:1211.6105

Dalcanton, J.J., Williams, B.F., Lang, D., et al., 2012, "The Panchromatic Hubble Andromeda Treasury," The Astrophysical Journal Supplements, 200, 18. arxiv:1204.0010

Lang, D. and Hogg, D.W., 2011, "Searching for comets on the World Wide Web: The orbit of 17P/Holmes from the behavior of photographers," The Astronomical Journal, 144, 46. arxiv:1103.6038

Lang, D., Hogg, D.W., Blanton, M. and Roweis, S., 2010, "Astrometry.net: Blind astrometric calibration of arbitrary astronomical images," The Astronomical Journal, 139, 1782. arxiv:0910.2233

Published, SDSS Survey Papers
Ahn, C.P. et al., 2013, "The Tenth Data Release of the Sloan Digital Sky Survey: First Spectroscopic Data from the SDSS-III Apache Point Observatory Galactic Evolution Experiment," Astrophysical Journal Supplements, arxiv:1307.7735

Eisenstein, D.J. et al., 2011, "SDSS-III: Massive Spectroscopic Surveys of the Distant Universe, the Milky Way Galaxy, and Extra-Solar Planetary Systems," The Astronomical Journal, 142, 72. arxiv:1101.1529

arXiv-only / unrefereed
Hogg, D. W. et al., 2013, "Maximizing Kepler science return per telemetered pixel: Detailed models of the focal plane in the two-wheel era," arxiv:1309.0653

Hogg, D.W., Bovy, J. and Lang, D., 2010, "Data analysis recipes: Fitting a model to data," arxiv:1008.4686

2015-06-03, University of Michigan, Local Group Astrostatistics,
"the Panchromatic Hubble Andromeda Treasury (PHAT) Survey: Astrometric Calibration" (source)

2015-03-31, University of Washington, Astronomy Seminar,
"Astronomical image reduction using the Tractor" (movie 1; source)

2015-02-11, Pasadena, WISE at 5 Conference,
"unWISE: unblurred coadds of the WISE images & photometry of 400,000,000 SDSS sources" (source)

2015-02-05, Western University, Physics & Astronomy Colloquium,
"The Web is My Observatory: Enabling science with heterogeneous images" (movie 1 movie 2; source)