Mapping the Mass: Successes and Challenges

Cluster Lensing: Peering into the Past, Preparing for the Future

Ian Dell’AntonioApril 15, 2013

• Major progress along the way• A few comments on shapes,shears and redshifts.• Where the “state of the art” is today• Where the Unsolved problems are.

Will focus only on the process of measuring the mass distribution—later “setting the stage” talks will deal with comparisons w/simulations and cosmological implications.

Setting the stage…goals

My (biased)view!

“...gravitational lens effects promise to furnish us withthe simplest and most accurate determination ofnebular masses[...] Until many plates of rich nebular fields taken under excellent conditions of seeing have been examined it would be dangerous, however to draw any conclusions[…]The chances for the successful application of this method grow rapidly with the size of the available telescope.”– ApJ, 86, 1937 pp 238, 245

“pre-history” of cluster lensing…

~25 years of cluster lensing…

1987: Soucail & collaborators report on arcs in Abell 370 and report that they must be due to gravitational lensing; Lynds and Petrosian report on survey of arcs in clusters—strong lensing is common in massive clusters.

1990: Tyson, Valdes \& Wenk– systematic alignment of arclets depends on radius from cluster center. Weak lensing mass profiles.

Richard et al. 2009

TVW 1990: Note the radius—not really “weak” lensing

From detection to mass

Miralda-Escudé (1991)--isothermal spheres Kaiser, Squires (1993): 2-d reconstructions

Mellier et al., Kneib et al. (1993) Multiple component modeling—introducing the standard algorithm.

Fahlman et al. (1994): Densitometry

Kaiser, Squires, Broadhurst (1995)—KSB method (later KSB+):

From mass to modelingSeitz &Schneider (1995): Indirect lensing potential

reconstruction, maximum likelihood

Kaiser, Schneider (1996): Aperture Mass statistic (filters modified over the years)

Smail et al., Colley et al. (1996): HST modeling of Strong lensing arcs.

(Khiabanian & ID, 2009)

Parametric versus Non-parametricNatarajan& Kneib (1996)—cluster substructure modeling.Natarajan&Kneib (1997), TKD (1998) –parametric modeling of cluster

strong lenses, using internal information of lensed galaxy structure.

Wittman et al. (2001) Weak lensing asa cluster finding tool:

Refregier & Bacon (2003) Shapelets—alternate ways of measuring shapes.

Improving the mapping

Inada et al. (2003): cluster-scale quasar lensesClowe et al. (2004): The Bullet Cluster; cluster

interactions/substructure

Goldberg&Bacon (2005): Detection of Flexion

Jullo et al. (2007) Bayesian MCMC methods applied to cluster strong lensing.

Coe et al. (2008) LensPerfect strong lensing modeler.

The current observational status

CLASH is underway—high resolution, multi-band imaging (photo-z); weak lensing from Subaru.

Deep legacy cluster fields to be observed with HSTMultiple strong+weak lensing surveys using CFHT, MMT,

Subaru underway. ~100 clusters with mapsNew generation cameras (HyperSuPrime, DECam) will

produce large samples (do we need to learn to deal with stacked clusters?).

JWST, EUCLID and LSST are now firmly on the horizon… this is a great time to be planning the future!

Weak lensing problems:PSF mappingGeneric problem– the galaxies are at positions (xg,yg), but the PSF is known only at

positions (x*,y*).Two problems: 1) how to derive PSF? From stars—either by measuring ellipticity

components of stars directly (a la KSB), or by modeling shapelet components (Massey&Refregier 2005).To transfer the measured components to the positions of the galaxies, traditionally low-ish order polynomial functions of x,y are used to model the components at the positions of the galaxies— an underconstrained problem. Two solutions:

1) PCA decompositions—find the components of PSF that persist over many images (Jarvis et al. 2002, Jee et al. 2007) and use the joint information to provide a denser sampling of the PSF

2) Use “calibration images” of starfields to span the space of PSF configurations with dense samplings, and then select the appropriate PSF based on the stars in the image (Rhodes et al. 2007, Schrabback et al 2010, Jee et al. 2007, Huwe et al. 2013).

(From Huwe et al. 2013)

Shape Measurement

KSB, KSB+ (still the standard)Lensfit (Miller et al. 2012), Stackfit (Jee 2012),

Multifit (Bosch et al. 2013?)—in principle better. But, more complicated and still catching up…

Testing the algorithms:

A vigorous program of simulations for weak lensing.STEP&STEP2: (Heymans et al. 2006, Massey et al. 2007)GREAT 08 (Bridle et al. 2009, 2010)GREAT 10 (Kitching et al. 2010, 2012)These have helped improve and validate PSF and galaxy shape

measurement.GREAT3 challenge—starts this summer!

(emulate for clusters?)

Shear Calibration:redshifts and tomographyLensing distortions depend on the angular diameter distance ratio.

Both a challenge and an opportunity.

Variations in background galaxy density need careful modeling (can bias substructure measures)

If the redshifts are known, this can give cosmological information (Tomography)

Projection.Lensing measures a projected mass distribution.The inferred mass (and distribution) depends on cluster orientation. Becker&Kravtsov (2011)—irreducible scatter in weak lensing due to cluster orientation of 20-30%.Can be modeled by including strong lensing, spectroscopic, and x-ray information.

Additional scatter due to LOS projections—this gets worse as the cluster mass gets lower. Hoekstra 2007 (~10-20% effect)—but can test and remove w/spectroscopic surveys (see Kurtz et al. 2012, Coe et al. 2012)

From Geller et al. 2010

State of the art: Quasar Strong lensesHST observations and modeling of quasar images, flux ratios, host galaxy shape, weak lensing profile, and time delay between A&B to constrain the mass profile (evidence for milli-lensing).

The challenge—time delays/followup observations are expensive—how many systems can be followed up?

Inada et al. 2006

Oguri et al. 2013

State of the Art: Strong lens modeling

(See also the presentations by Medezinski, Jee,Richard).10s of arc positions and morphologies with spectroscopic

redshifts!

• Parametric lens modeling using using Mass-follows-light models, NFW, etc…)

• Non-parametric modeling—should there be a bake-off?

State of the art: cluster weak lensing

Reconstruction via KSB+: still effective (see GREAT challenges).

Minimum entropy reconstructions of potentialSPH-inspired “particle lensing” (Deb et al. 2011)

Ultimately will want to model the distortion in a LENSfit/MULTIFIT way. Can we make it work?

State of the art: substructure

Weak lensing:Okabe et al. 2013—weak lensing in

Coma to detect minihalos; Huwe et al. 2013—weak lensing substructure via filtered aperture mass method.

Strong lensing: Modeling of cluster galaxy potentials with galaxies mature—but how dependent are we on the baryonic properties of galaxies?

Weak lensing non-parametric (Huwe et al. 2013)

State of the art: Clusters and LSSClusters are embedded in the large scale structure. But the mass density is lower so first “detections” were only via cross-correlation spectroscopically selected groups (Geller et al. 2005) or galaxy distributions (Gray et al. 2002).This is changing (see Jauzac et al. 2013). What can we learn from the connection between clusters and the LSS they are embedded in?

Gray et al. 2002

Challenges (a partial list)

• Strong lens modeling—parametric vs. non-parametric. Are both necessary?• How settled is the question of the “shallow” cores from strong lensing?

• Lensing-selected cluster samples—less efficient than X-rays or MaxBCG-like algorithms; are they still relevant?

• PSF distortions in weak (and strong?) lensing: now that we are going after very small mass and small scale substructure, we will need new ways to model the small-scale PSF variations. What data/modeling technique will take us there?

• How well do shape measurement techniques work in the medium shear regime? How well do they deal with crowding? (Call for another level of “GREAT” challenge?)

Challenges (a partial list)

• Faint cluster galaxies--is there a reliable way to remove them from lensing analysis without throwing out many of the

• Projection effects currently dealt with via spectroscopy. This is expensive! Is there another way?

• What is the right way to make use of other (X-ray, spectroscopic, S-Z, etc…) data?

• Others?