ELTs for Cluster Cosmology

ELTs for Cluster Cosmology Anja von der Linden Stony Brook University UCLA, April 4th, 2018

Cosmology only ~5% of the Universe in a form we understand (stars, planets, atoms baryons ) what is dark energy? cosmological constant: single parameter (Λ) equation of state: w = p/ c 2 1 consistent with all observations no satisfying theoretical explanation time-varying scalar field: bw 6= 1 modified gravity: change General Relativity on large scales precision measurements of w are the goal of most cosmology experiments

Cosmology only ~5% of the Universe in a form we understand (stars, planets, atoms baryons ) what is dark matter? is it cold? is it collisionless? what is the mass of neutrinos? why are neutrino masses so small? from oscillations: X m & 0.056 ev cosmological constraints on mass could inform planning next-generation ground-based neutrino experiments

Clusters as probes of the Dark Universe Dark Energy Task Force (2006): clusters are messy... since then: - improved statistical techniques - low-scatter mass proxies (X-rays) - weak-lensing mass calibration - competitive constraints from X-ray, optical, and mm surveys Cosmic Visions Report (2016): The number of massive galaxy clusters could emerge as the most powerful cosmological probe if the masses of the clusters can be accurately measured.

Cluster cosmology: state of the art Weighing the Giants alone places 15% constraint on w; one of the tightest single-probe constraints today WtG IV, Mantz, AvdL et al. 2015 WtG based on only(!) ~200 X-ray-selected (ROSAT), most massive clusters at z<0.5 50 with Subaru weak-lensing masses 90 with Chandra imaging constraints also from optical and SZ cluster surveys; constraints from Dark Energy Survey (DES) clusters imminent! next decade: 10 000s of clusters, multiple selection methods (optical, SZ, X-ray), to z~2 + LSST, Euclid, WFIRST weak lensing tremendous potential

Beyond w neutrino masses: σ 8 0.70 0.75 0.80 0.85 0.90 vanilla + Ω k + w 0 + N eff + r WtG IV, Mantz, AvdL et al. 2015 0.0 0.1 0.2 0.3 0.4 0.5 0.6 m ν (ev) informative constraints on neutrino masses even when other parameters (e.g. w) allowed to vary f(r) gravity: Cataneo et al. 2014

Statistical potential for cluster cosmology Cosmic Visions Report (2016): The number of massive galaxy clusters could emerge as the most powerful cosmological probe if the masses of the clusters can be accurately measured. How can ELTs help cluster cosmology? statistical errors only http://www.wfirst-hls-cosmology.org Krause & Eifler 2016

How can ELTs help cluster cosmology? Calibration: - photometric redshifts of weak-lensing source galaxies - deblending of weak-lensing source galaxies (?) High-z clusters: - confirmation + spectroscopic redshifts - redshift-dependent w models Individual cluster analyses: - shear ratio test (both in strong lensing + weak lensing) - kinematic weak lensing - cluster profiles + mass mapping (+ lensed SNe) Key technology: - high-multiplex MOS over several arcminutes, optical + NIR

The Dark Universe with Clusters and ELTs Comprehensive deep imaging + spectroscopy in cluster fields! We find interesting things when we look at clusters! Great synergy with other applications: - galaxy evolution in high-density environments - cosmic telescopes: magnify the high-redshift universe - lensed transients -...

Talk outline I. Ingredients for cluster cosmology - Counting clusters - cluster surveys in the 2020s - ELTs: confirmation + spec-z of high-z clusters II. Mass calibration - accurate weak lensing masses - ELTs: photo-z calibration III. Individual cluster analyses - kinematic weak lensing - shear ratio test

Counting Halos halo mass function number of gravitationally bound halos sensitive to cosmological model both geometry (volume) and growth of structure (evolution of mass function) idea: AvdL; movie: Matt Becker, Ralf Kaehler, Yao-Yuan Mao, Rachel Reddick, Risa Wechsler (Stanford/SLAC)

Ingredients for cluster counts cosmology 1. prediction for halo mass function 2. cluster survey (X-rays, SZ, optical) with well understood selection function X-ray optical 3. relation between survey observable and cluster mass 4. self-consistent statistical framework SZ log luminosity 1.5 0.5 0.5 1.0 1.5 1.0 0.5 0.0 0.5 1.0 log mass cosmology

SZ (2) Finding clusters X-rays X-ray SZ X-rays: thermal bremsstrahlung from Intra- Cluster Medium (ICM) millimeter: Sunyaev-Zeldovich effect - inverse Compton scattering of CMB photons on ICM optical: galaxy population - overdensity of (red) galaxies optical

SZ Finding clusters optical / NIR highest completeness, to relatively low masses - subject to projection effects - red sequence finding works very well at z 1, but RS not well populated at higher redshifts X-rays: in principle, very high purity and completeness (every extended extragalactic source is a cluster) - in practice: limited angular resolution leads to impurity / incompleteness due to AGN confusion - large scatter Lx - mass of ~40% Sunyaev-Zeldovich effect: nearly redshift-independent mass selection threshold high purity and completeness relatively small scatter in SZ signal - mass of ~20% SZ is the best cluster finder esp. at high redshifts X-rays optical

Cluster Surveys in the 2020s many surveys in optical, SZ, X-rays on-going, starting, or planned great synergy prospects; 100s of 1000s of clusters key developments: mass calibration + redshift leverage best cosmology constraints will come from the combination from SZ multi-wavelength datasets DES optical+nir LSST SPTpol, SPT-3G, CMB-S4 Euclid X-rays WFIRST X-rays erosita (Advanced) ACTpol Planck SZ

Stage-III cluster surveys: SPT-3G: SZ, deep, small area (2500 ) - detects clusters to ~10 14 M SZ and X-ray cluster catalogs AdvACT: SZ, shallow, wide area within LSST footprint (15000 ) - detects massive (rare) clusters erosita: X-rays, full sky, mainly low-z; gas masses for a subset Benson et al. 2014 By LSST start, SZ surveys will find: large fraction of clusters with >10 14 M within SPT-3G footprint (almost all of) the most massive, rare clusters in the south Mass-limited samples to z~2 - interesting targets for ELTs!

SZ Future SZ surveys: Simons Observatory + CMB-S4 Simons Observatory: ~LSST area, similar noise as SPT-3G (but smaller dish, ~6m?), starting ~2021 CMB-S4: full-sky (?), lower noise than SPT-3G (6m?), 2022 X-rays CMB-S4 Science Book thin lines: CMB-S4 mass thresholds (50% completeness) for 3, 2, 1 resolution lower resolution compromises z 1 cluster finding + CMB lensing mass calibration

SZ Synergy: SZ cluster surveys + ELT follow-up SZ surveys are our best cluster finders at high redshifts... but they cannot measure cluster redshifts X-rays At z 1.3: can get red sequence redshifts from LSST At z 1.3? Euclid/WFIRST - bright objects ELTs: - MOS over several arcmin; for z>1.5, need NIR ~3.5 Newman et al. 2015 MOO J1142+1527, z=1.2: Spitzer + SZ Gonzalez et al. 2015

(3) Measuring cluster masses survey observables (optical richness, X-ray luminosity, SZ decrement) do not measure cluster mass directly correlate with mass, but with considerable scatter: 20-40% X-ray is it necessary to measure the mass of every cluster in the survey? no, we only need to know how to describe the population P ( observable M ) Cluster mass here: same as in simulations that predict halo mass function usually 3D overdensity mass: M = 4 3 c (z)r 3

Mass-Observable Relations simplest assumption: the mean of the observable follows a power-law relation with mass X-ray hobsi = M intrinsic scatter around the mean is log-normal motivation: assumption of self-similarity (Kaiser 1984) scatter + survey selection

Importance of the mass normalization 10 5 Ω M =0.25, Ω Λ =0.75, h =0.72 Vikhlinin et al. 2009b Rozo et al. 2010 N(>M), h 3 Mpc 3 10 6 10 7 10 8 10 9 z =0.025 0.25 z =0.55 0.90 10 14 10 15 M 500, h 1 M for σ8 (+ neutrino masses, etc.) already current results limited by systematic uncertainty in mass normalization currently ~5%, LSST-era surveys will require ~ 1-2%

Low-scatter mass proxies follow-up X-ray observations can provide a number of low-scatter ( 10%) mass proxies: ICM temperature TX; gas mass Mgas; YX = Mgas x TX Mantz et al. 2016 essential for measuring shape and scatter of M-O relation do not provide absolute mass calibration

Absolute Masses? need observables that can be related to the gravitational potential: X-ray hydrostatic masses non-thermal bias, TX calibration galaxy dynamics large scatter and bias weak lensing + small bias: accurate - scatter ~30%

(Cluster) (Optical) (Weak) Lensing mass deflects light measure light deflection to estimate cluster mass sensitive to total mass 2d masses: no assumption on dynamical state needed strong lensing: multiple images, arcs probes cluster core weak lensing (shear-based): statistical tangential alignment probes mass on large scales each background galaxy unbiased, noisy estimator of local deflection (shear) comes for free with weak-lensing surveys

Large Synoptic Survey Telescope (LSST) 8.4m (~6.7m effective) telescope, 10 sq. deg. camera images the entire visible sky every 3 nights 10-year survey of 20000 sq. deg. in 6 bands data from commissioning camera starting 2020; full camera 2021 main survey starts 2023 LSST is built for weak lensing: high-quality shear measurements + photo-z s out to z~1.4 LSST will deliver weak-lensing masses for all southern clusters SZ Vast statistical power have to understand the systematics DESC: LSST Dark Energy Science Collaboration

Ingredients for cluster mass measurements Shear induced on background galaxy depends on: cluster mass (distribution) redshift WtG 1 To measure cluster mass, need 1. reduced shear measurements 2. redshifts / redshift distribution 3. (some) assumption on mass distribution... and need to understand the systematics of each! WtG 1I

(i) Shear measurements bias in shear estimates bias in cluster mass estimate requirements inherited from cosmic shear requirements + need to calibrate to (only) ~1%, cf. ~0.01% for cosmic shear cluster-specific issues: - shear in clusters is larger - dense fields: blending, background subtraction blending shear bias - ELT high-resolution, wide-field imaging: empirically map shear bias from blending

shear on background galaxy depends on redshift shear(z) is a steep function right behind the cluster, then flattens out error in mass from photo-z s depends on cluster redshift (ii) Photometric Redshifts LSST weak lensing catalog: redshift distribution of resolved galaxies peaks at z~0.8 good photo-z s require coverage of 4000Å break; LSST z 1.4 Euclid / WFIRST: NIR photometry higher-redshift range ELTs: spectroscopic training samples wide-field MOS, high multiplexing Jeff Newman s talk

p(z) s - redshift probability distributions WtG III: Applegate, AvdL et al. 2014 Fractional Mass Bias within 1.5 Mpc 0.15 0.10 0.05 0.00 0.05 Point Estimators P(z) Method 0.2 0.3 0.4 0.5 0.6 0.7 Cluster Redshift state-of-the art analyses use p(z)s : redshift probability distributions training / calibration: need to map p(z ugrizy)

Redshifts cluster-specific concern: dilution by cluster members cluster galaxies not sheared (and no empirical evidence for intrinsic alignments, e.g. Sifon et al. 2015) any contamination of lensing sample causes mass underestimate what are the properties of cluster galaxies? At LSST depth? At z~1? In typical clusters? Is there dust in clusters that can bias the p(z) of background galaxies? ELTs: spectroscopic training samples in cluster fields wide-field ( 5 ) MOS, high multiplexing for z>1 clusters: need NIR for background galaxies

(iii) Mass model lensing sensitive to all mass along line-of-sight measures projected 2D masses for relation to halo mass function, need to infer 3D mass galaxies are intrinsically elliptical weak lensing is noisy can typically measure only one parameter reliably fit spherically symmetric profile (also breaks mass-sheet degeneracy) projected mass depends on cluster triaxiality / orientation / substructure, structure along LOS e.g. Meneghetti et al. 2010, Hoekstra 2003, 2011 (3D) lensing masses have intrinsic, irreducible scatter of 20% (ground-based: scatter from shape noise also ~20% total scatter: ~30%) (e.g. Becker & Kravtsov 2011)

(iii) Mass model Is the average lensing mass (un-)biased calibratable? methodology can be well tested on simulations NFW profile good description only to virial radius (Becker&Kravtsov 11) need to quantify mass bias as function of mass, radius, redshift, fitting method, miscentering, cosmology,... and include baryons mass bias small (a few percent), but needs to be accurately calibrated for cluster cosmology

1. prediction for halo mass function 2. cluster survey (4) Statistical model 3. mass-observable relation 4. self-consistent statistical framework: WtG IV simultaneously fit for cluster masses, mass-observable relation, and cosmology WtG: low-scatter X-ray mass proxies + individual cluster lensing masses can measure intrinsic scatter in lensing masses: ln M WL =0.17 ± 0.06 WtG V

(5) Avoiding confirmation bias Klein & Roodman 2005 Croft & Daly 2011: of 28 measurements of Ω Λ, only 2 are more than 1σ from the WMAP results need to blind analyses to avoid confirmation bias requires extensive testing - builds confidence that results are reliable state of the art cluster cosmology results are blinded: WtG, DES

Cool Things with Individual Clusters

(i) Kinematic Weak Lensing Weak lensing is noisy... because of intrinsic ellipticities of galaxies Precision on the mass of a single cluster limited by the number of galaxies in the lensing analysis shear or inclination? (How) can we do better? Can we tell the intrinsic shape of the galaxy? Potential pay-offs: Smaller statistical uncertainty per cluster Can measure more parameters per cluster: concentration, halo shape and orientation, density profile... reduce intrinsic scatter?

Kinematic Weak Lensing Slide by Eric Huff

Kinematic Weak Lensing Tully-Fisher relation: tight relation between rotation velocity and luminosity of spiral galaxies Huff et al. 2013: Use measured rotation velocity to estimate inclination angle disentangle shear and inclination! can significantly reduce intrinsic scatter in WL observations Key technology: (slit-mask) multiobject spectroscopy over wide FOV

(ii) Shear Ratio Test ratio of shear measured at different redshifts sensitive to geometry of Universe, especially w mass model (+uncertainties) drop out to first order strong lensing: need multiple image families at different z s weak lensing: need lots of really good redshifts Taylor et al. 2007 Auger et al.

Shear Ratio Test 34 multiple image families; A1689 (Jullo et al. 2009) stacked signal from 25 clusters, weak lensing + 5-filter photo-z s (Kelly, AvdL et al. 2014) 1.5 = 0.7 t(x)/ t(x = 1) 1.0 0.5 0.0 0 1 2 3 4 5 x =! s /! l ELTs: strong lensing: identification + spec-z s of multiply imaged systems weak lensing: spec-z s for background galaxies

(iii) Dynamical masses? velocity dispersions, Jeans modeling, caustics... various assumptions on symmetry / virialization mass bias and scatter both large; depend on method, number / population of galaxies targeted most robust methods: >100 objects; to large radii Old et al. 2014, 2015

SZ + Dark Matter! cluster cosmology will generate high-quality datasets for lots of other science, including studying dark matter: density profiles: cores from Self-Interacting Dark Matter? But: need to understand baryonic feedback! X-rays merging clusters: probes of (astro-)physics under extreme conditions, probe of Self- Interacting Dark Matter with monitoring: multiply imaged transients test dark matter model

Summary exciting time for cluster cosmology! multiple surveys in 2020s: optical, SZ, X-rays need to measure mass-observable relation: mean relation + shape and size of scatter relative mass calibration: low-scatter mass proxies absolute mass calibration: weak lensing, LSST +Euclid/WFIRST potential ELT contributions: - spec-z training samples - confirmation / spec-z of high-redshift clusters unique applications: - kinematic weak lensing: reduce weak lensing noise - shear ratio test: ELT key capability: wide-field, high-multiplexing multi-object spectroscopy SZ