LensPerfect Analysis of Abell 1689

The Highest Resolution Mass Map of Galaxy Cluster Substructure To Date Without Assuming Light Traces Mass

Press release at STScI

My own release, including other interesting tidbits

paper: Coe et al. 2010, ApJ 723,1678 (arXiv)
-- multiple image stamps (17M)
-- delensed multiple image stamps (8M)
Resource paper on dark matter profiles: (Coe 2010, astro-ph/1005.0411)

Comments welcome via e-mail or at scirate.com (free & easy registration): LPA1689, DMprofiles

dark matter contours

Substructures ~25 kpc across (roughly the visible size of galaxies) on average are revealed (contours) within the central ~400 kpc (diameter) of this galaxy cluster core, thanks to an analysis of strong gravitational lensing: 135 multiple images of 42 galaxies (and a total of 168 images of 55 knots within those galaxies) are identified. Their positions (pink squares above; labeled below) are used to constrain the distribution of cluster mass (mostly dark matter) interior to them (white outline above).

Multiple images (click to enlarge)

The gravity of A1689 bends the light of more distant galaxies according to Einstein's General Relativity. We see these galaxies distorted and displaced, as if through a glass lens. Based on these displacements, we can determine the strength of this lumpy lens as a function of position across its surface. It's like a giant puzzle: We must figure out where all the mass must be in the core of A1689 to produce the lensed images where and how we see them.

Our model successfully "de-lenses" the multiple images so we may see the galaxies as they truly look:
(click to enlarge)

Observed images are in the top row and de-lensed images are in the bottom row. Note the de-lensed images are all look roughly the same now,* including clumps oriented the same way, and are at the same position (see coordinate axes).

*(Note clumps 0,1,2 all align in the delensed images of 10a & 10b. Clump #10a3 appears to belong to a different galaxy. And image #10c has been demagnified, so we don't resolve all the clumps in that image. Colors in the #10c images appear slightly different as they were produced by a different method after modeling and subtracting the cluster galaxies.)

Our LensPerfect (Coe08) mass model de-lenses all images back to the same "source" position, down to the individual clumps ...for all 168 images of 55 clumps within 42 lensed galaxies. No other analysis method can accomplish this. Models of other methods fail to reproduce the observed multiple image positions by 50 pixels or more. That's ~50x the observational uncertainties of ~1 pixel. Other methods do well (even perfectly) by stopping at ~30 images. However by modeling only ~1/4 the images, they sacrifice ~4x the mass map resolution (~2x along each axis).

Some of the other methods assume that the visible light traces the distribution of the mass, including most of the dark matter. We instead map out the mass (mostly dark matter) without making this assumption. We can then test how well the light does indeed trace the mass. Our mass model shows that the light does trace the mass pretty well, at least qualitatively. However there are some deviations which we will be investigating more thoroughly. The mass-to-light ratios of individual galaxies will also be interesting to measure.

We stress that it can be very useful and informative to assume that light traces mass. This assumption probably is a good one more often than not. Ultimately we would like to develop a new method which assumes that light approximately traces mass, using the light distribution as a prior on the mass distribution.

A main scientific result from Coe10 verifies what others have found: the mass of A1689 is more centrally concentrated than we expect based on simulations of dark matter structure formation. (For an explanation, see my DMprofiles paper.) Some other clusters appear to be similarly over-concentrated. More definitive results are expected from CLASH, a Multi-Cycle Hubble Treasury project. If clusters are indeed over-concentrated, one explanation would be "Early Dark Energy": ~10x more at early times (e.g., Ω_DE ~ 0.10 at z~6) than expected from ΛCDM (assuming Einstein's cosmological constant).