Issue 
A&A
Volume 594, October 2016
Planck 2015 results



Article Number  A24  
Number of page(s)  19  
Section  Cosmology (including clusters of galaxies)  
DOI  https://doi.org/10.1051/00046361/201525833  
Published online  20 September 2016 
Planck 2015 results
XXIV. Cosmology from SunyaevZeldovich cluster counts
^{1} APC, AstroParticule et Cosmologie, Université Paris Diderot, CNRS/IN2P3, CEA/lrfu, Observatoire de Paris, Sorbonne Paris Cité, 10 rue Alice Domon et Léonie Duquet, 75205 Paris Cedex 13, France
^{2} Aalto University Metsähovi Radio Observatory and Dept of Radio Science and Engineering, PO Box 13000, 00076 Aalto, Finland
^{3} African Institute for Mathematical Sciences, 68 Melrose Road, Muizenberg 7945, Cape Town, South Africa
^{4} Agenzia Spaziale Italiana Science Data Center, via del Politecnico snc, 00133 Roma, Italy
^{5} Aix Marseille Université, CNRS, LAM (Laboratoire d’Astrophysique de Marseille) UMR 7326, 13388 Marseille, France
^{6} Astrophysics Group, Cavendish Laboratory, University of Cambridge, J J Thomson Avenue, Cambridge CB3 0HE, UK
^{7} Astrophysics & Cosmology Research Unit, School of Mathematics, Statistics & Computer Science, University of KwaZuluNatal, Westville Campus, Private Bag X54001, 4000 South Africa
^{8} Atacama Large Millimeter/submillimeter Array, ALMA Santiago Central Offices, Alonso de Cordova 3107, Vitacura, 763 0355 Casilla, Santiago, Chile
^{9} CGEE, SCS Qd 9, Lote C, Torre C, 4° andar, Ed. Parque Cidade Corporate, CEP 70308200, Brasília, DF, Brazil
^{10} CITA, University of Toronto, 60 St. George St., Toronto, ON M5S 3H8, Canada
^{11} CNRS, IRAP, 9 Av. colonel Roche, BP 44346, 31028 Toulouse Cedex 4, France
^{12} CRANN, Trinity College, Dublin, Ireland
^{13} California Institute of Technology, Pasadena, California, CA 91125, USA
^{14} Centre for Theoretical Cosmology, DAMTP, University of Cambridge, Wilberforce Road, Cambridge CB3 0WA, UK
^{15} Centro de Estudios de Física del Cosmos de Aragón (CEFCA), Plaza San Juan 1, planta 2, 44001 Teruel, Spain
^{16} Computational Cosmology Center, Lawrence Berkeley National Laboratory, Berkeley, California, CA 94720, USA
^{17} Consejo Superior de Investigaciones Científicas (CSIC), 28006 Madrid, Spain
^{18} DSM/Irfu/SPP, CEASaclay, 91191 GifsurYvette Cedex, France
^{19} DTU Space, National Space Institute, Technical University of Denmark, Elektrovej 327, 2800 Kgs. Lyngby, Denmark
^{20} Département de Physique Théorique, Université de Genève, 24 Quai E. Ansermet, 1211 Genève 4, Switzerland
^{21} Departamento de Astrofísica, Universidad de La Laguna (ULL), 38206 La Laguna, Tenerife, Spain
^{22} Departamento de Física, Universidad de Oviedo, Avda. Calvo Sotelo s/n, 33007 Oviedo, Spain
^{23} Department of Astronomy and Astrophysics, University of Toronto, 50 Saint George Street, 38100 Toronto, Ontario, Canada
^{24} Department of Astrophysics/IMAPP, Radboud University Nijmegen, PO Box 9010, 6500 GL Nijmegen, The Netherlands
^{25} Department of Physics & Astronomy, University of British Columbia, 6224 Agricultural Road, Vancouver, British Columbia, Canada
^{26} Department of Physics and Astronomy, Dana and David Dornsife College of Letter, Arts and Sciences, University of Southern California, Los Angeles, CA 90089, USA
^{27} Department of Physics and Astronomy, University College London, London WC1E 6BT, UK
^{28} Department of Physics, Florida State University, Keen Physics Building, 77 Chieftan Way, Tallahassee, Florida, USA
^{29} Department of Physics, Gustaf Hällströmin katu 2a, University of Helsinki, 00014 Helsinki, Finland
^{30} Department of Physics, Princeton University, Princeton, New Jersey, NJ 08544, USA
^{31} Department of Physics, University of California, Santa Barbara, California, CA 93106, USA
^{32} Department of Physics, University of Illinois at UrbanaChampaign, 1110 West Green Street, Urbana, Illinois, USA
^{33} Dipartimento di Fisica e Astronomia G. Galilei, Università degli Studi di Padova, via Marzolo 8, 35131 Padova, Italy
^{34} Dipartimento di Fisica e Scienze della Terra, Università di Ferrara, via Saragat 1, 44122 Ferrara, Italy
^{35} Dipartimento di Fisica, Università La Sapienza, P.le A. Moro 2, 00133 Roma, Italy
^{36} Dipartimento di Fisica, Università degli Studi di Milano, via Celoria 16, 20133 Milano, Italy
^{37} Dipartimento di Fisica, Università degli Studi di Trieste, via A. Valerio 2, 34127 Trieste, Italy
^{38} Dipartimento di Matematica, Università di Roma Tor Vergata, via della Ricerca Scientifica 1, 00133 Roma, Italy
^{39} Discovery Center, Niels Bohr Institute, Blegdamsvej 17, Copenhagen, Denmark
^{40} Discovery Center, Niels Bohr Institute, Copenhagen University, Blegdamsvej 17, 1165 Copenhagen, Denmark
^{41} European Southern Observatory, ESO Vitacura, Alonso de Cordova 3107, Vitacura, Casilla 19001, Santiago, Chile
^{42} European Space Agency, ESAC, Planck Science Office, Camino bajo del Castillo, s/n, Urbanización Villafranca del Castillo, 28691 Villanueva de la Cañada, Madrid, Spain
^{43} European Space Agency, ESTEC, Keplerlaan 1, 2201 AZ Noordwijk, The Netherlands
^{44} Gran Sasso Science Institute, INFN, viale F. Crispi 7, 67100 L’ Aquila, Italy
^{45} HGSFP and University of Heidelberg, Theoretical Physics Department, Philosophenweg 16, 69120 Heidelberg, Germany
^{46} Haverford College Astronomy Department, 370 Lancaster Avenue, Haverford, Pennsylvania, USA
^{47} Helsinki Institute of Physics, Gustaf Hällströmin katu 2, University of Helsinki, 00560 Helsinki, Finland
^{48} INAF–Osservatorio Astrofisico di Catania, via S. Sofia 78, Catania, Italy
^{49} INAF–Osservatorio Astronomico di Padova, Vicolo dell’Osservatorio 5, 35131 Padova, Italy
^{50} INAF–Osservatorio Astronomico di Roma, via di Frascati 33, 00040 Monte Porzio Catone, Italy
^{51} INAF–Osservatorio Astronomico di Trieste, via G.B. Tiepolo 11, 40127 Trieste, Italy
^{52} INAF/IASF Bologna, via Gobetti 101, 40129 Bologna, Italy
^{53} INAF/IASF Milano, via E. Bassini 15, 20133 Milano, Italy
^{54} INFN, Sezione di Bologna, via Irnerio 46, 40126, Bologna, Italy
^{55} INFN, Sezione di Roma 1, Università di Roma Sapienza, Piazzale Aldo Moro 2, 00185 Roma, Italy
^{56} INFN, Sezione di Roma 2, Università di Roma TorVergata, via della Ricerca Scientifica 1, 00185 Roma, Italy
^{57} INFN/National Institute for Nuclear Physics, via Valerio 2, 34127 Trieste, Italy
^{58} IPAG: Institut de Planétologie et d’Astrophysique de Grenoble, Université Grenoble Alpes, IPAG, CNRS, IPAG, 38000 Grenoble, France
^{59} ISDC, Department of Astronomy, University of Geneva, ch. d’Ecogia 16, 1290 Versoix, Switzerland
^{60} IUCAA, Post Bag 4, Ganeshkhind, Pune University Campus, 411 007 Pune, India
^{61} Imperial College London, Astrophysics group, Blackett Laboratory, Prince Consort Road, London, SW7 2AZ, UK
^{62} Infrared Processing and Analysis Center, California Institute of Technology, Pasadena, CA 91125, USA
^{63} Institut Néel, CNRS, Université Joseph Fourier Grenoble I, 25 rue des Martyrs, Grenoble, France
^{64} Institut Universitaire de France, 103 Bd SaintMichel, 75005 Paris, France
^{65} Institut d’Astrophysique Spatiale, CNRS, Univ. ParisSud, Université ParisSaclay, Bât. 121, 91405 Orsay Cedex, France
^{66} Institut d’Astrophysique de Paris, CNRS (UMR7095), 98bis Boulevard Arago, 75014 Paris, France
^{67} Institut für Theoretische Teilchenphysik und Kosmologie, RWTH Aachen University, 52056 Aachen, Germany
^{68} Institute for Space Sciences, BucharestMagurale, Romania
^{69} Institute of Astronomy, University of Cambridge, Madingley Road, Cambridge CB3 0HA, UK
^{70} Institute of Theoretical Astrophysics, University of Oslo, Blindern, 0371 Oslo, Norway
^{71} Instituto de Astrofísica de Canarias, C/Vía Láctea s/n, 38205 La Laguna, Tenerife, Spain
^{72} Instituto de Física de Cantabria (CSICUniversidad de Cantabria), Avda. de los Castros s/n, 39005 Santander, Spain
^{73} Istituto Nazionale di Fisica Nucleare, Sezione di Padova, via Marzolo 8, 35131 Padova, Italy
^{74} Jet Propulsion Laboratory, California Institute of Technology, 4800 Oak Grove Drive, Pasadena, California, USA
^{75} Jodrell Bank Centre for Astrophysics, Alan Turing Building, School of Physics and Astronomy, The University of Manchester, Oxford Road, Manchester, M13 9PL, UK
^{76} Kavli Institute for Cosmological Physics, University of Chicago, Chicago, IL 60637, USA
^{77} Kavli Institute for Cosmology Cambridge, Madingley Road, Cambridge, CB3 0HA, UK
^{78} Kazan Federal University, 18 Kremlyovskaya St., 420008 Kazan, Russia
^{79} LAL, Université ParisSud, CNRS/IN2P3, 91898 Orsay, France
^{80} LERMA, CNRS, Observatoire de Paris, 61 Avenue de l’Observatoire, 75000 Paris, France
^{81} Laboratoire AIM, IRFU/Service d’Astrophysique – CEA/DSM – CNRS – Université Paris Diderot, Bât. 709, CEASaclay, 91191 GifsurYvette Cedex, France
^{82} Laboratoire Traitement et Communication de l’Information, CNRS (UMR 5141) and Télécom ParisTech, 46 rue Barrault, 75634 Paris Cedex 13, France
^{83} Laboratoire de Physique Subatomique et Cosmologie, Université GrenobleAlpes, CNRS/IN2P3, 53 rue des Martyrs, 38026 Grenoble Cedex, France
^{84} Laboratoire de Physique Théorique, Université ParisSud 11 & CNRS, Bâtiment 210, 91405 Orsay, France
^{85} Lawrence Berkeley National Laboratory, Berkeley, California, USA
^{86} Lebedev Physical Institute of the Russian Academy of Sciences, Astro Space Centre, 84/32 Profsoyuznaya st., GS7, 117997 Moscow, Russia
^{87} MaxPlanckInstitut für Astrophysik, KarlSchwarzschildStr. 1, 85741 Garching, Germany
^{88} McGill Physics, Ernest Rutherford Physics Building, McGill University, 3600 rue University, Montréal, QC, H3A 2T8, Canada
^{89} National University of Ireland, Department of Experimental Physics, Maynooth, Co. Kildare, Ireland
^{90} Nicolaus Copernicus Astronomical Center, Bartycka 18, 00716 Warsaw, Poland
^{91} Niels Bohr Institute, Blegdamsvej 17, 1165 Copenhagen, Denmark
^{92} Niels Bohr Institute, Copenhagen University, Blegdamsvej 17, 1165 Copenhagen, Denmark
^{93} Nordita (Nordic Institute for Theoretical Physics), Roslagstullsbacken 23, 106 91 Stockholm, Sweden
^{94} Optical Science Laboratory, University College London, Gower Street, London, UK
^{95} SISSA, Astrophysics Sector, via Bonomea 265, 34136 Trieste, Italy
^{96} SMARTEST Research Centre, Università degli Studi eCampus, via Isimbardi 10, 22060 Novedrate (CO), Italy
^{97} School of Physics and Astronomy, Cardiff University, Queens Buildings, The Parade, Cardiff, CF24 3AA, UK
^{98} School of Physics and Astronomy, University of Nottingham, Nottingham NG7 2RD, UK
^{99} Sorbonne UniversitéUPMC, UMR7095, Institut d’Astrophysique de Paris, 98bis Boulevard Arago, 75014, Paris, France
^{100} Space Research Institute (IKI), Russian Academy of Sciences, Profsoyuznaya Str, 84/32, 117997 Moscow, Russia
^{101} Space Sciences Laboratory, University of California, Berkeley, California, CA 94720, USA
^{102} Special Astrophysical Observatory, Russian Academy of Sciences, Nizhnij Arkhyz, Zelenchukskiy region, 369167 KarachaiCherkessian Republic, Russia
^{103} Stanford University, Dept of Physics, Varian Physics Bldg, 382 via Pueblo Mall, Stanford, California, CA 943054060, USA
^{104} SubDepartment of Astrophysics, University of Oxford, Keble Road, Oxford OX1 3RH, UK
^{105} The Oskar Klein Centre for Cosmoparticle Physics, Department of Physics, Stockholm University, AlbaNova, 106 91 Stockholm, Sweden
^{106} Theory Division, PHTH, CERN, 1211 Geneva 23, Switzerland
^{107} UPMC Univ. Paris 06, UMR7095, 98bis Boulevard Arago, 75014 Paris, France
^{108} Université de Toulouse, UPSOMP, IRAP, 31028 Toulouse Cedex 4, France
^{109} University Observatory, Ludwig Maximilian University of Munich, Scheinerstrasse 1, 81679 Munich, Germany
^{110} University of Granada, Departamento de Física Teórica y del Cosmos, Facultad de Ciencias, 81679 Granada, Spain
^{111} University of Granada, Instituto Carlos I de Física Teórica y Computacional, Granada, Spain
^{112} Warsaw University Observatory, Aleje Ujazdowskie 4, 00478 Warszawa, Poland
^{⋆}
Corresponding authors: A. Bonaldi, email: anna.bonaldi@manchester.ac.uk; M. Roman, email: matthieu.roman@apc.univparis7.fr
Received: 6 February 2015
Accepted: 14 June 2016
We present cluster counts and corresponding cosmological constraints from the Planck full mission data set. Our catalogue consists of 439 clusters detected via their SunyaevZeldovich (SZ) signal down to a signaltonoise ratio of 6, and is more than a factor of 2 larger than the 2013 Planck cluster cosmology sample. The counts are consistent with those from 2013 and yield compatible constraints under the same modelling assumptions. Taking advantage of the larger catalogue, we extend our analysis to the twodimensional distribution in redshift and signaltonoise. We use mass estimates from two recent studies of gravitational lensing of background galaxies by Planck clusters to provide priors on the hydrostatic bias parameter, (1−b). In addition, we use lensing of cosmic microwave background (CMB) temperature fluctuations by Planck clusters as an independent constraint on this parameter. These various calibrations imply constraints on the presentday amplitude of matter fluctuations in varying degrees of tension with those from the Planck analysis of primary fluctuations in the CMB; for the lowest estimated values of (1−b) the tension is mild, only a little over one standard deviation, while it remains substantial (3.7σ) for the largest estimated value. We also examine constraints on extensions to the base flat ΛCDM model by combining the cluster and CMB constraints. The combination appears to favour nonminimal neutrino masses, but this possibility does little to relieve the overall tension because it simultaneously lowers the implied value of the Hubble parameter, thereby exacerbating the discrepancy with most current astrophysical estimates. Improving the precision of cluster mass calibrations from the current 10%level to 1% would significantly strengthen these combined analyses and provide a stringent test of the base ΛCDM model.
Key words: cosmological parameters / largescale structure of Universe
© ESO, 2016
1. Introduction
Galaxy cluster counts are a standard cosmological tool that has found powerful application in recent SunyaevZeldovich (SZ) surveys performed by the Atacama Cosmology Telescope (ACT, Swetz et al. 2011; Hasselfield et al. 2013), the South Pole Telescope (SPT, Carlstrom et al. 2011; Benson et al. 2013; Reichardt et al. 2013; Bocquet et al. 2015), and the Planck satellite^{1} (Tauber et al. 2010; Planck Collaboration I 2011). The abundance of clusters and its evolution are sensitive to the cosmic matter density, Ω_{m}, and the present amplitude of density fluctuations, characterized by σ_{8}, the rms linear overdensity in spheres of radius 8h^{1} Mpc. The primary cosmic microwave background (CMB) anisotropies, on the other hand, reflect the density perturbation power spectrum at the time of recombination. This difference is important because a comparison of the amplitude of the perturbations at the two epochs tests the evolution of density perturbations from recombination until today, enabling us to look for possible extensions to the base Λcold dark matter (CDM) model, such as nonminimal neutrino masses or nonzero curvature.
Launched on 14 May 2009, Planck scanned the entire sky twice a year from 12 August 2009 to 23 October 2013, at angular resolutions from 33′ to 5′ with two instruments: the Low Frequency Instrument (LFI; Bersanelli et al. 2010; Mennella et al. 2011), covering bands centred at 30, 44, and 70 GHz, and the High Frequency Instrument (HFI; Lamarre et al. 2010; Planck HFI Core Team 2011), covering bands centred at 100, 143, 217, 353, 545, and 857 GHz.
An initial set of cosmology results appeared in 2013, based on the first 15.5 months of data (Planck Collaboration I 2014), including cosmological constraints from the redshift distribution of 189 galaxy clusters detected at signaltonoise (S/N) > 7 (hereafter, our “first analysis” or the “2013 analysis”, Planck Collaboration XX 2014). The present paper is part of the second set of cosmology results obtained from the full mission data set; it is based on an updated cluster sample introduced in an accompanying paper (the PSZ2, Planck Collaboration I 2016).
Our first analysis found fewer clusters than predicted by Planck’s base ΛCDM model, expressed as a tension between the cluster constraints on (Ω_{m},σ_{8}) and those from the primary CMB anisotropies (Planck Collaboration XVI 2014). This could reflect the need for an extension to the base ΛCDM model or indicate that clusters are more massive than determined by the SZ signalmass scaling relation adopted in 2013.
The cluster mass scale is the largest source of uncertainty in interpretation of the cluster counts. We based our first analysis on Xray mass proxies that rely on the assumption of hydrostatic equilibrium. Simulations demonstrate that this assumption can be violated by bulk motions in the gas or by nonthermal sources of pressure (e.g., magnetic fields or cosmic rays, Nagai et al. 2007; Piffaretti & Valdarnini 2008; Meneghetti et al. 2010). Systematics in the Xray analyses (e.g., instrument calibration, temperature structure in the gas) could also bias the mass measurements significantly (Rasia et al. 2006, 2012). We quantified our ignorance of the true mass scale of clusters with a mass bias parameter that was varied over the range 0−30%, with a baseline value of 20% (see below for the definition of the mass bias), as suggested by numerical simulations (see the Appendix of Planck Collaboration XX 2014).
Gravitational lensing studies of the SZ signalmass relation are particularly valuable in this context because they are independent of the dynamical state of the cluster (Marrone et al. 2012; Planck Collaboration Int. III 2013), although they also, of course, can be affected by systematic effects (e.g., Becker & Kravtsov 2011). New, more precise lensing mass measurements for Planck clusters have appeared since our 2013 analysis (von der Linden et al. 2014b; Hoekstra et al. 2015). We incorporate these new results as prior constraints on the mass bias in the present analysis. Two other improvements over 2013 are the use of a larger cluster catalogue and analysis of the counts in signaltonoise as well as redshift.
In addition, we apply a novel method to measure cluster masses through lensing of the CMB anisotropies. This method, presented in Melin & Bartlett (2015), enables us to use Planck data alone to constrain the cluster mass scale. It provides an important independent mass determination, which we compare to the galaxy lensing results, and one that is representative in the sense that it averages over the entire cluster cosmology sample, rather than a particularly chosen subsample. It is, however, less well tested than the other lensing methods because of its novel nature, and we comment on various potential systematics deserving further examination.
Our conventions throughout the paper are as follows. We specify cluster mass, M_{500}, as the total mass within a sphere of radius R_{500}, defined as the radius within which the mean mass overdensity of the cluster is 500 times the cosmic critical density at its redshift, z: , with ρ_{c}(z) = 3H^{2}(z) / (8πG), where H(z) is the Hubble parameter with presentday value H_{0} = h × 100 km s^{1} Mpc^{1}. We give SZ signal strength, Y_{500}, in terms of the Compton yprofile integrated within a sphere of radius R_{500}, and we assume that all clusters follow the universal pressure profile of Arnaud et al. (2010). Density parameters are defined relative to the presentday critical density, e.g., Ω_{m} = ρ_{m}/ρ_{c}(z = 0) for the total matter density, ρ_{m}.
We begin in the next section with a presentation of the Planck 2015 cluster cosmology samples. In Sect. 3 we develop our model for the cluster counts in both redshift and signaltonoise, including a discussion of the scaling relation, the scatter and the sample selection function. Section 4 examines the overall cluster mass scale in light of recent gravitational lensing measurements; we also present our own calibration of the cluster mass scale based on lensing of the CMB temperature fluctuations. Construction of the cluster likelihood and selection of external data sets is detailed in Sect. 5. We compare results based on our new likelihood to the 2013 Planck cluster results in Sect. 6. We then present our 2015 cosmological constraints in Sect. 7, summarizing and discussing the results in Sect. 8. We examine the potential impact of different modelling uncertainties in the Appendix.
Fig. 1 Massredshift distribution of the Planck cosmological samples colourcoded by their signaltonoise, q. The baseline MMF3 2015 cosmological sample is shown as the small filled circles. Objects which were in the MMF3 2013 cosmological sample are marked by crosses, while those in the 2015 intersection sample are shown as open circles. The final samples are defined by q> 6. The mass M_{Yz} is the Planck mass proxy (see text; Arnaud et al., in prep.). 
2. The Planck cosmological samples
We detect clusters across the six highest frequency Planck bands (100−857 GHz, Planck Collaboration VII 2016; Planck Collaboration VIII 2016) using two implementations of the multifrequency matched filter (MMF3 and MMF1, Melin et al. 2006; Planck Collaboration XXIX 2014) and a Bayesian extension (PwS, Carvalho et al. 2009) that all incorporate the known (nonrelativistic) SZ spectral signature and a model for the spatial profile of the signal. The latter is taken to be the socalled “universal pressure profile” from Arnaud et al. (2010) – with the nonstandard selfsimilar scaling – and parameterized by an angular scale, θ_{500}.
We empirically characterize noise (all nonSZ signals) in localized sky patches (10deg on a side for MMF3) using the set of crossfrequency powerspectra. We construct the filters with the resulting noise weights, and we then filter the set of six frequency maps over a range of cluster scales, θ_{500}, spanning 1–35 arcmin. The filter returns an estimate of Y_{500} for each scale, based on the adopted profile template, and sources are finally assigned the θ_{500} (and hence Y_{500}) value of the scale that maximizes their signaltonoise. Details are given in Planck Collaboration XXIX (2014) and in an accompanying paper introducing the Planck fullmission SZ catalogue (PSZ2, Planck Collaboration XXVII 2016).
We define two cosmological samples from the general PSZ2 catalogues, one consisting of detections by the MMF3 matched filter and the other of objects detected by all three methods (the intersection catalogue). Both are defined by a signaltonoise (denoted q throughout) cut of q> 6. We then apply a mask to remove regions of high dust emission and point sources, leaving 65% of the sky unmasked. The general catalogues, noise maps and masks can be downloaded from the Planck Legacy Archive^{2}.
The cosmological samples can be easily constructed from the PSZ2 union and MMF3 catalogues. The MMF3 cosmology sample is the subsample of the MMF3 catalogue defined by q> 6 and for which the entry in the union catalogue has COSMO=“T”. The intersection cosmology sample is defined from the union catalogue by the criteria COSMO=“T”, PIPEDET=111, and q> 6.
Figure 1 shows the distribution of these samples in mass and redshift, together with the 2013 cosmology sample. The mass here is the Planck mass proxy, M_{Yz}, defined in Arnaud et al. (in prep.) (see also Sect. 7.2.2. in Planck Collaboration XXIX 2014) and taken from the PSZ2 catalogue. It is calculated using the Planck sizeflux posterior contours in conjunction with Xray priors to break the sizeflux degeneracy inherent to the large Planck beams (see, e.g., Fig. 16 of Planck Collaboration XXVII 2016). The samples span masses in the range (2−10) × 10^{14}M_{⊙} and redshifts from z = 0 to 1^{3}. This quantityM_{Yz} is used in the external lensing mass calibration measurements, as discussed in Sect. 4.
The MMF3 (intersection) sample contains 439 (493) detections. We note that the intersection catalogue has more objects than the MMF3 catalogue because of the different definitions of the signaltonoise in the various catalogues. The signaltonoise for the intersection catalogue corresponds to the highest signaltonoise of the three detection algorithms (MMF1, MMF3, or PwS), while for the MMF3 catalogue we use its corresponding signaltonoise. As a consequence, the lowest value for the MMF3 signaltonoise in the intersection sample is 4.8. We note that, while being above our detection limit, the Virgo and the Perseus clusters are not part of our samples. This is because Virgo is too extended to be blindly detected by our algorithms and Perseus is close to a masked region.
The 2015 MMF3 cosmology sample contains all but one of the 189 clusters of the 2013 MMF3 sample. The missing cluster is PSZ1 980, which falls inside the 2015 point source mask. Six (14) redshifts are missing from the MMF3 (intersection) sample. Our analysis accounts for these by renormalizing the observed counts to redistribute the missing fraction uniformly across the considered redshift range [0,1]. The small number of clusters with missing redshifts has no significant impact on our results.
We use the MMF3 cosmology sample at q> 6 for our baseline analysis and the intersection sample for consistency checks, as detailed in the Appendix. In particular, we show that the intersection sample yields equivalent constraints.
3. Modelling cluster counts
From the theoretical perspective, cluster abundance is a function of halo mass and redshift, as specified by the mass function. Observationally, we detect clusters in Planck through their SZ signal strength or, equivalently, their signaltonoise and measure their redshift with followup observations. The observed cluster counts are therefore a function of redshift, z, and signaltonoise, q. While we restricted our 2013 cosmology analysis to the redshift distribution alone (Planck Collaboration XX 2014), the larger catalogue afforded by the full mission data set offers the possibility of an analysis in both redshift and signaltonoise. We therefore develop the theory in terms of the joint distribution of clusters in the (z,q)plane and then relate it to the more specific analysis of the redshift distribution to compare with our previous results.
3.1. Counts as a function of redshift and signaltonoise
The distribution of clusters in redshift and and signaltonoise can be written as (1)with (2)i.e., the dark matter halo mass function times the volume element. We adopt the mass function from Tinker et al. (2008) throughout, apart from the Appendix, where we compare to the Watson et al. (2013a) mass function as a test of modelling robustness; there, we show that the Watson et al. (2013a) mass function yields constraints similar to those from the Tinker et al. (2008) mass function, but shifted by about 1σ towards higher Ω_{m} and lower σ_{8} along the main degeneracy line.
The quantity is the distribution of q given the mean signaltonoise value, , predicted by the model for a cluster of mass M_{500} and redshift z located at Galactic coordinates (l,b)^{4}. This latter quantity is defined as the ratio of the mean SZ signal expected of a cluster, , as given in Eq. (7), and the detection filter noise, : (3)The filter noise depends on sky location (l,b) and the cluster angular size, , which introduces additional dependence on mass and redshift. More detail on σ_{f} can be found in Planck Collaboration XX (2014) (see in particular Fig. 4 therein).
The distribution incorporates noise fluctuations and intrinsic scatter in the actual cluster Y_{500} around the mean value, , predicted from the scaling relation. We discuss this scaling relation and our lognormal model for the intrinsic scatter below, and Sect. 4 examines the calibration of the overall mass scale for the scaling relation.
The redshift distribution of clusters detected at q>q_{cat} is the integral of Eq. (1) over signaltonoise, (4)with (5)Equation (4) is equivalent to the expression used in our 2013 analysis if we write it in the form (6)where χ(Y_{500},θ_{500},l,b) is the survey selection function at q>q_{cat} in terms of true cluster parameters (Sect. 3.3), and P(lnY_{500},θ_{500}  z,M_{500}) is the distribution of these parameters given cluster mass and redshift. We specify the relation between Eqs. (5) and (6) in the next section.
3.2. Observablemass relations
A crucial element of our modelling is the relation between cluster observables, Y_{500} and θ_{500}, and halo mass and redshift. Due to intrinsic variations in cluster properties, this relation is described by a distribution function, P(lnY_{500},θ_{500}  M_{500},z), whose mean values are specified by the scaling relations and .
We use the same form for these scaling relations as in our 2013 analysis: (7)and (8)where θ_{∗} = 6.997 arcmin, and fiducial ranges for the parameters Y_{∗}, α, and β are listed in Table 1; these values are identical to those used in our 2013 analysis. Unless otherwise stated, we use Gaussian distributions with mean and standard deviation given by these values as prior constraints; one notable exception will be when we simultaneously fit for α and cosmological parameters. In the above expressions, D_{A}(z) is the angular diameter distance and E(z) ≡ H(z) /H_{0}.
These scaling relations have been established by Xray observations, as detailed in the Appendix of Planck Collaboration XX 2014, and rely on mass determinations, M_{X}, based on hydrostatic equilibrium of the intracluster gas. The “mass bias” parameter, b, assumed to be constant in both mass and redshift, allows for any difference between the Xray determined masses and true cluster halo mass: M_{X} = (1−b)M_{500}. This is discussed at length in Sect. 4.
We adopt a lognormal^{5} distribution for Y_{500} around its mean value , and a delta function for θ_{500} centred on : (9)where and are given by Eqs. (7) and (8). The δfunction maintains the empirical definition of R_{500} that is used in observational determinations of the profile.
We can now specify the relation between Eqs. (5) and (6) by noting that (10)where P [q  q_{m}] is the distribution of observed signaltonoise, q, given the model value, q_{m}. The second distribution represents intrinsic cluster scatter, which we write in terms of our observablemass distribution, Eq. (9), as (11)Performing the integral of Eq. (5), we find (12)with the definition of our survey selection function (13)We then reproduce Eq. (6) by using the first line of Eqs. (11) and (3).
3.3. Selection function and survey completeness
The fundamental quantity describing the survey selection is P [q  q_{m}], introduced in Eq. (10). It gives the observed signaltonoise, used to select SZ sources, as a function of model (“true”) cluster parameters through q_{m}(Y_{500},θ_{500},l,b), and it defines the “survey selection function” χ(Y_{500},θ_{500},l,b) via Eq. (13). We characterize the survey selection in two ways. The first is with an analytical model and the second employs a Monte Carlo extraction of simulated sources injected into the Planck maps. In addition, we perform an external validation of our selection function using known Xray clusters.
The analytical model assumes pure Gaussian noise, in which case we simply have . The survey selection function is then given by the error function (and we refer to this as the ERF completeness function), (14)This model can be applied to a catalogue with welldefined noise properties (i.e., σ_{f}), such as our MMF3 catalogue, but not to the intersection catalogue based on the simultaneous detection with three different methods. This is our motivation for choosing the MMF3 catalogue as our baseline.
In the Monte Carlo approach, we inject simulated clusters directly into the Planck maps and (re)extract them with the complete detection pipeline. Details are given in the accompanying 2015 SZ catalogue paper, Planck Collaboration XXVII (2016). This method provides a more comprehensive description of the survey selection by accounting for a variety of effects beyond noise. In particular, we vary the shape of the SZ profile at fixed Y_{500} and θ_{500} to quantify its effect on catalogue completeness. The difference between the MonteCarlo and ERF completeness results in a change in modelled number counts of typically ~ 2.5% (with a maximum of 9%) in each redshift bin.
We also perform an external check of the survey completeness using known Xray clusters from the Meta Catalogue of Xray Clusters (MCXC) compilation (Piffaretti et al. 2011) and also SPT clusters from Bleem et al. (2015). Details are given in the 2015 SZ catalogue paper, Planck Collaboration XXVII (2016). For the MCXC compilation, we rely on the expectation that at redshifts z< 0.2 any Planckdetected cluster should be found in one of the ROSAT catalogues (Chamballu et al. 2012) because at low redshift ROSAT probes to lower masses than Planck^{6}. The MCXC catalogue provides a truth table, replacing the input cluster list of the simulations, and we compute completeness as the ratio of objects in the cosmology catalogue to the total number of clusters. As discussed in Planck Collaboration XXVII (2016), the results are consistent with Gaussian noise and bound the possible effect of profile variations. We arrive at the same conclusion when applying the technique to the SPT catalogue.
Planck Collaboration XXVII (2016) discusses completeness checks in greater detail. One possible source of bias is the presence of correlated IR emission from cluster member galaxies. Planck Collaboration XXIII (2016) suggests that IR point sources may contribute significantly to the cluster SED at the Planck frequencies, especially at higher redshift. The potential impact of this effect warrants further study in future work.
We thus have different estimations of the selection function for MMF3 and the intersection catalogues. We test the sensitivity of our cosmological constraints to the selection function in Appendix A by comparing results obtained with the different methods and catalogues. We find that our results are insensitive to the choice of completeness model (Fig A.1), and we therefore adopt the analytical ERF completeness function for simplicity throughout the paper.
4. The cluster mass scale
The characteristic mass scale of our cluster sample is the critical element in our analysis of the counts. It is controlled by the mass bias factor, 1−b, accounting for any difference between the Xray mass proxies used to establish the scaling relations and the true (halo) mass: M_{X} = (1−b)M_{500}. Such a difference could arise from cluster physics (such as a violation of hydrostatic equilibrium or temperature structure in the gas, Rasia et al. 2006, 2012, 2014), from observational effects (e.g., instrumental calibration), or from selection effects biasing the Xray samples relative to SZ or massselected samples (Angulo et al. 2012).
In our 2013 analysis, we adopted a flat prior on the mass bias over the range 1−b = [0.7,1.0], with a reference model defined by 1−b = 0.8. This was motivated by a comparison of the Y−M_{X} relation with published Y−M relations derived from numerical simulations, as detailed in the Appendix of Planck Collaboration XX (2014); this estimate was consistent with most (although not all) predictions for any violation of hydrostatic equilibrium, as well as observational constraints from the available lensing observations. Effects other than cluster physics can contribute to the mass bias, as discussed in our earlier paper, and as emphasized by the survey of cluster multiband scaling relations by Rozo et al. (2014a,b,c).
The mass bias was the largest uncertainty in our 2013 analysis, and it severely hampered understanding of the tension found between constraints from the primary CMB and the cluster counts. Here, we incorporate new lensing mass determinations of Planck clusters to constrain the mass bias. We also apply a novel method to measure object masses based on lensing of CMB temperature anisotropies behind clusters (Melin & Bartlett 2015). These constraints are used as prior information in our analysis of the counts. As we will see, however, uncertainty in the mass bias remains our largest source of uncertainty, mainly because these various determinations continue to differ by up to 30%.
In general, the mass bias could depend on cluster mass and redshift, although we will model it by a constant in the following. Our motivation is one of practicality: the limited size and precision of current lensing samples makes it difficult to constrain any more than a constant value, i.e., the overall mass scale of our catalogue. Large lensing surveys like Euclid, WFIRST, and the Large Synoptic Survey Telescope, as well as CMB lensing, will improve this situation in coming years.
Summary of mass scale priors.
4.1. Constraints from gravitational shear
Several cluster samples with high quality gravitational shear mass measurements have appeared since 2013. Among these, the Weighing the Giants (WtG, von der Linden et al. 2014a), CLASH (Postman et al. 2012; Merten et al. 2015; Umetsu et al. 2014), and Canadian Cluster Comparison Project (CCCP, Hoekstra et al. 2015) programmes offer constraints on our mass bias factor, 1−b, through direct comparison of the lensing masses to the Planck mass proxy, M_{Yz}.
The analysis by the WtG programme of 22 clusters from the 2013 Planck cosmology sample yields 1−b = 0.688 ± 0.072. Their result lies at the very extreme of the range explored in Planck Collaboration XX (2014) and would substantially reduce the tension found between primary CMB and galaxy cluster constraints. Hoekstra et al. (2015) report a smaller bias of 1−b = 0.78 ± 0.07 (stat) ± 0.06 (sys) for a set of 20 common clusters, which is in good agreement with the fiducial value adopted in our 2013 analysis. In our new analysis we add the statistical and systematic uncertainties in quadrature (see Table 2).
The two samples overlap, but not completely, and, as discussed in detail by Hoekstra et al. (2015), there are numerous differences between the two analyses. These include treatment of source redshifts, contamination by cluster members and methods of extracting a mass estimate from the lensing data. And while the two mass calibrations differ in a way that attracts particular attention in the present context, they are statistically consistent, separated by about one standard deviation.
4.2. Constraints from CMB lensing
Measuring cluster mass through CMB lensing (Lewis & Challinor 2006) has been discussed in the literature for some time since the study performed by Zaldarriaga & Seljak (1999). We apply a new technique for measuring cluster masses through lensing of CMB temperature anisotropies (Melin & Bartlett 2015), allowing us to calibrate the scaling relations using only Planck data. This is a valuable alternative to the galaxy lensing observations because it is independent and affected by different possible systematics. Additionally, we can apply it to the entire cluster sample to obtain a mass calibration representative of an SZ fluxselected sample. Similar approaches using CMB lensing to measure halo masses were recently applied by SPT (Baxter et al. 2015) and ACT (Madhavacheril et al. 2015).
Our method first extracts a clean CMB temperature map with a constrained internal linear combination (ILC) of the Planck frequency channels in the region around each cluster; the ILC is constrained to nullify the SZ signal from the clusters themselves and provide a clean CMB map of 5 arcmin resolution. Using a quadratic estimator on the CMB map, we reconstruct the lensing potential in the field and then filter it to obtain an estimate of the cluster mass. The filter is an NFW profile (Navarro et al. 1997) with scale radius set by the Planck mass proxy for each cluster, and designed to return an estimate of the ratio M_{lens}/M_{Yz}, where M_{Yz} is the Planck SZ mass proxy. These individual measurements are corrected for any meanfield bias by subtracting identical filter measurements on blank fields; this accounts for effects of apodization over the cluster fields and correlated noise. The technique has been tested on realistic simulations of Planck frequency maps. More detail can be found in Melin & Bartlett (2015).
Fig. 2 Cluster mass scale determined by CMB lensing. We show the ratio of cluster lensing mass, M_{lens}, to the SZ mass proxy, M_{Yz}, as a function of the mass proxy for clusters in the MMF3 2015 cosmology sample. The cluster mass is measured through lensing of CMB temperature anisotropies in the Planck data (Melin & Bartlett 2015). Individual mass measurements have low signaltonoise, but we determine a mean ratio for the sample of M_{lens}/M_{Yz} = 1 / (1−b) = 0.99 ± 0.19. For clarity, only some of the error bars are plotted (see text). 
Figure 2 shows M_{lens}/M_{Yz} as a function of M_{Yz} for all clusters in the MMF3 cosmology sample. Each point is an individual cluster^{7}. For clarity, only some of the error bars on the ratio are shown; the error bars vary from 1.8 at the high mass end to 8.5 at the low mass end, with a median of 4.2. There is no indication of a correlation between the ratio and M_{Yz}, and we therefore fit for a constant ratio of M_{lens}/M_{Yz} by taking the weighted mean (using the individual measurement uncertainties as provided by the filter) over the full data set. If the ratio differs from unity, we apply a correction to account for the fact that our filter aperture was not perfectly matched to the clusters. The correction is calculated assuming an NFW profile and is of the order of one percent.
The final result is 1 / (1−b) = 0.99 ± 0.19, traced by the blue band in the figure. We note that the method constrains 1 / (1−b) rather than 1−b as in the case of the shear measurements. The calculated uncertainty on the weighted mean is consistent with a bootstrap analysis, where we create new catalogues of the same size as the original by sampling objects from the full catalogue with replacement; the uncertainty from the bootstrap is then taken as the standard deviation of the bootstrap means.
The uncertainty 0.19 is statistical. Melin & Bartlett (2015) quote an uncertainty of 0.28 for the 62 ESZXMM clusters based on simulations including only tSZ, kSZ, primary CMB and instrumental noise. Scaling this number to 433 (number of objects with a redshift in the cosmological sample) gives , in broad agreement with our value of 0.19. The difference can likely be attributed to the fact that the Planck cosmological sample is on average less massive than the ESZXMM sample, and that the Planck maps are more complex than the model adopted in Melin & Bartlett (2015).
We have obtained a 5σ measurement of the sample mass scale using CMB lensing. We emphasize, however, that the method is new and under development. A number of potential systematic effects require further study, including cluster miscentring and mismatch between filter shape and actual cluster profiles; these would tend to reduce the observed masses from their true values. On the other hand, effects such as contributions from mass correlated on large scales with the clusters (e.g., filaments or neighbouring halos) and contamination by infrared and radio sources could increase the observed signal. We are examining these issues in a study of the ESZXMM sample that will be published at a later date.
4.3. Summary
The three mass bias priors are summarized in Table 2, and we will extract cosmological constraints from each one. We favour these three lensing results because of their direct comparison to the Planck mass proxy. We will assume Gaussian distributions for 1−b (gravitational shear) or 1 / (1−b) (CMB lensing), with standard deviations given by the error column. We adopt the CCCP mass calibration as our baseline, and give the CMB lensing result less weight in our interpretation because of its novelty and the ongoing studies of the issues mentioned in the previous section.
5. Analysis methodology
5.1. Likelihood
Our 2013 analysis employed a likelihood built on the cluster redshift distribution, dN/ dz. With the larger 2015 catalogue, our baseline likelihood is now constructed on counts in the (z,q)plane. We divide the catalogue into bins of size Δz = 0.1 (10 bins) and Δlog q = 0.25 (5 bins), each with an observed number N(z_{i},q_{j}) = N_{ij} of clusters. Modelling the observed counts, N_{ij}, as independent Poisson random variables, our loglikelihood is (15)where N_{z} and N_{q} are the total number of redshift and signaltonoise bins, respectively. The mean number of objects in each bin is predicted by theory according to Eq. (1): (16)which depends on the cosmological (and cluster modelling) parameters. In practice, we use a Monte Carlo Markov chain (MCMC) to map the likelihood surface around the maximum and establish confidence limits.
Equation (15) assumes that the bins are uncorrelated, while a more complete description would include correlations due to largescale clustering. In practice, our cluster sample contains mostly high mass systems for which the impact of these effects is weak (e.g., Hu & Kravtsov 2003, in particular their Fig. 4 for the impact on constraints in the (Ω_{m},σ_{8}) plane).
5.2. External data sets
Cluster counts cannot constrain all pertinent cosmological parameters; they are most sensitive to Ω_{m} and σ_{8}, and when analysing the counts alone we must apply additional observational constraints as priors on other parameters. For this purpose, we adopt Big Bang nucleosynthesis (BBN) constraints from Steigman (2008), Ω_{b}h^{2} = 0.022 ± 0.002 (for a recent review on BBN, see Olive 2013), and constraints from baryon acoustic oscillations (BAO). The latter combine the 6dF Galaxy Survey (Beutler et al. 2011), the SDSS Main Galaxy Sample (Padmanabhan et al. 2012; Anderson et al. 2012) and the BOSS DR11 (Anderson et al. 2014). We refer the reader to Sect. 5.2 in Planck Collaboration XIII (2016) for details of the combination. We also include a prior on n_{s} from Planck Collaboration XVI (2014), n_{s} = 0.9624 ± 0.014. When explicitly specified in the text, we add the supernovæ constraint from SNLSII and SNLS3: the Joint Lightcurve Analysis constraint (JLA, Betoule et al. 2014). The BAO are particularly sensitive to H_{0}, while the supernovæ allow precise constraints on the dark energy equationofstate parameter, w.
6. Comparison to 2013
We begin by verifying consistency with the results of Planck Collaboration XX (2014) (Sect. 6.1) based on the onedimensional likelihood over the redshift distribution, dN/ dz (Eq. (4)). We then examine the effect of changing to the full twodimensional likelihood, dN/ dzdq (Eq. (1)) in Sect. 6.2. For this purpose we compare constraints on the total matter density, Ω_{m}, and the lineartheory amplitude of the density perturbations today, σ_{8}, using the cluster counts in combination with external data and fixing the mass bias. The twodimensional likelihood dN/ dzdq is then adopted as the baseline in the rest of the paper.
6.1. Constraints on Ω_{m} and σ_{8}: onedimensional analysis
Figure 3 presents constraints from the MMF3 cluster counts combined with the BAO and BBN priors of Sect. 5.2; we refer to this data combination as “SZ+BAO+BBN”. To compare to results from our 2013 analysis (the grey, filled ellipses), we use a onedimensional likelihood based on Eq. (4) over the redshift distribution and have adopted the reference scaling relation of 2013, i.e., Eqs. (7) and (8), with the mass bias value fixed to 1−b = 0.8. For the present comparison, we use the updated BAO constraints discussed in Sect. 5.2; these are stronger than the BAO constraints used in the 2013 analysis, and the grey contours shown here are consequently smaller than in Planck Collaboration XX (2014).
Limiting the 2015 catalogue to q> 8.5 produces a sample with 190 clusters, similar to the 2013 cosmology catalogue (189 objects). The two sets of constraints demonstrate good consistency, and they remain consistent while becoming tighter as we decrease the signaltonoise threshold of the 2015 catalogue. Under similar assumptions, our 2015 analysis thus confirms the 2013 results reported in Planck Collaboration XX (2014).
The area of the ellipse from q = 8.5 to q = 6 decreases by a factor of 1.3. This is substantially less than the factor of 2.3 expected from the ratio of the number of objects in the two samples. The difference may be related to the decreasing goodnessoffit of the best model as the signaltonoise decreases. When incorporated, the uncertainty on the mass calibration 1−b will also restrict the reduction of the ellipse area.
Figure 4 overlays the observed cluster redshift distribution on the predictions from the bestfit model in each case. We see that the models do not match the counts in the second and third redshift bins (counting from z = 0), and that the discrepancy, already marginally present at the high signaltonoise cut corresponding to the 2013 catalogue, becomes more pronounced towards the lower signaltonoise thresholds. This discrepancy cannot be attributed to redshift errors in the first bins because the majority of the redshifts are spectroscopic and the size of the bins is large (Δ_{z} = 0.1); for example, the first two redshift bins contain 208 clusters, of which 200 have spectroscopic redshifts. The dependence on signaltonoise may suggest that the data prefer a different slope, α, of the scaling relation than allowed by the prior of Table 1. We explore the effect of relaxing the Xray prior on α in the next section.
Fig. 3 Contours at 95% for different signaltonoise thresholds, q = 8.5, 7, and 6, applied to the 2015 MMF3 cosmology sample for the SZ+BAO+BBN data set. The contours are compatible with the 2013 constraints (Planck Collaboration XX 2014), shown as the filled, light grey ellipses at 68 and 95% (for the BAO and BBN priors of Sect 5.2; see text). The 2015 catalogue thresholded at q> 8.5 has a similar number of clusters (190) as the 2013 catalogue (189). This comparison is made using the analytical errorfunction model for completeness and adopts the reference observablemass scaling relation of the 2013 analysis (1−b = 0.8, see text). The redshift distributions of the bestfit models are shown in Fig. 4. For this figure and Fig. 4, we use the onedimensional likelihood over the redshift distribution, dN/ dz (Eq. (4)). 
Fig. 4 Comparison of observed counts (points with error bars) with predictions of the bestfit models (solid lines) from the onedimensional likelihood for three different thresholds applied to the 2015 MMF3 cosmology sample. The mismatch between observed and predicted counts in the second and third lowest redshift bins, already noticed in the 2013 analysis, increases at lower thresholds, q. The bestfit models are defined by the constraints shown in Fig. 3. For this figure and Fig. 3, we use our onedimensional likelihood over the redshift distribution, dN/ dz (Eq. (4)), with the mass biased fixed at (1−b) = 0.8. 
6.2. Constraints on Ω_{m} and σ_{8}: twodimensional analysis
In Fig. 5 we compare constraints from the one and twodimensional likelihood with α either free or with the prior of Table 1. For this comparison, we continue with the “SZ+BAO+BBN” data set, but adopt the CCCP prior for the mass bias and only consider the full 2015 MMF3 catalogue at q> 6.
The grey and black contours and lines in Fig. 5 show results from the onedimensional likelihood fit to the redshift distribution using, respectively, the Xray prior on α and leaving α free. The redshift counts do indeed prefer a steeper slope, with a posterior of α = 2.23 ± 0.18 in the latter case and a shift of the constraints along their degeneracy ridges. To explore this preference, we split the analysis into low and high redshift bin sets divided at z = 0.2, finding that neither the high nor the low redshift bin set prefers the steeper slope by itself; it appears only when analyzing all the bins. As described further in Appendix A, there is a subtle interplay between parameters that is masked by the degeneracies and difficult to interpret with the present data set.
A related issue is the acceptability of the model fit. We define a generalized χ^{2} measure of goodnessoffit as , determining the probability to exceed (PTE) the observed value using Monte Carlo simulations of Poisson statistics for each bin with the bestfit model mean . The observed value of the fit drops from 17 (PTE = 0.07) with the Xray prior, to 15 (PTE = 0.11) when leaving α free. When leaving α free, Ω_{m} increases and σ_{8} decreases, following their correlation with α shown by the contours, and their uncertainty increases due to the added parameter.
The twodimensional likelihood over dN/ dzdq better constrains the slope when α is free, as shown by the violet curves and contours. In this case, the preferred value drops back towards the Xray prior: α = 1.89 ± 0.11, just over 1σ from the central Xray value. Reimposing the Xray prior on α with the twodimensional likelihood (blue curves) does little to change the parameter constraints. Although the onedimensional likelihood prefers a steeper slope than the Xray prior, the twodimensional analysis does not, and the cosmological constraints remain robust to varying α.
We define a generalized χ^{2} statistic as described above, now over the twodimensional bins in the (z,q)plane. This generalized χ^{2} for the fit with the Xray prior is 43 (PTE = 0.28), compared to χ^{2} = 45 (PTE = 0.23) when α is a free parameter.
Figure 6 displays the redshift distribution of the bestfit models in all four cases. Despite their apparent difficulty in matching the second and third redshift bins, the PTE values suggest that these fits are moderately good to acceptable. We note that, as mentioned briefly in Sect. 5.1, clustering effects will increase the scatter in each bin slightly over the Poisson value we have assumed, causing our quoted PTE values to be somewhat smaller than the true ones.
Fig. 5 Comparison of constraints from the onedimensional (dN/ dz) and twodimensional (dN/ dzdq) likelihoods on cosmological parameters and the scaling relation mass exponent, α. This comparison uses the MMF3 catalogue, the CCCP prior on the mass bias and the SZ+BAO+BBN data set. The corresponding bestfit model redshift distributions are shown in Fig. 6. 
Fig. 6 Redshift distribution of bestfit models from the four analysis cases shown in Fig. 5. The observed counts in the MMF3 catalogue (q> 6) are plotted as the red points with error bars, and as in Fig. 5 we adopt the CCCP mass prior with the SZ+BAO+BBN data set. 
7. Cosmological constraints 2015
We extract constraints on Ω_{m} and σ_{8} from the cluster counts in combination with external data, imposing the different cluster mass scale calibrations as prior distributions on the mass bias. In Sect. 7.1, we compare our new constraints to and then combine them with those from the CMB anisotropies in the base ΛCDM model. We study parameter extensions to the base model in Sect. 7.2. In the following, we adopt as our baseline the 2015 twodimensional SZ likelihood with the CCCP mass bias prior, α free and β = 2 / 3 fixed in Eq. (7). All quoted intervals are 68% confidence and all upper/lower limits are 95% confidence.
7.1. Base ΛCDM
Summary of Planck 2015 cluster cosmology constraints
7.1.1. Constraints on Ω_{m} and σ_{8}: comparison to primary CMB parameters
Fig. 7 Comparison of constraints from the CMB to those from the cluster counts in the (Ω_{m},σ_{8})plane. The green, blue and violet contours give the cluster constraints (twodimensional likelihood) at 68 and 95% for the WtG, CCCP, and CMB lensing mass calibrations, respectively, as listed in Table 2. These constraints are obtained from the MMF3 catalogue with the SZ+BAO+BBN data set and α free (hence the SZα notation). Constraints from the Planck TT, TE, EE+lowP CMB likelihood (hereafter, Planck primary CMB) are shown as the dashed contours enclosing 68 and 95% confidence regions (Planck Collaboration XIII 2016), while the grey shaded region also includes BAO. The red contours give results from a joint analysis of the cluster counts and the Planck lensing power spectrum (Planck Collaboration XV 2016), adopting our external priors on n_{s} and Ω_{b}h^{2} with the mass bias parameter free and α constrained by the Xray prior (hence the SZ notation without the subscript α). 
Our 2013 analysis brought to light tension between constraints on Ω_{m} and σ_{8} from the cluster counts and those from the primary CMB in the base ΛCDM model. In that analysis, we adopted a flat prior on the mass bias over the range 1−b = [0.7,1.0], with a reference model defined by 1−b = 0.8 (see discussion in the Appendix of Planck Collaboration XX 2014). Given the good consistency between the 2013 and 2015 cluster results (Fig. 3), we expect the tension to remain under the same assumptions concerning the mass bias.
Figure 7 compares our 2015 cluster constraints (MMF3 SZ+BAO+BBN) to those for the base ΛCDM model from the Planck CMB anisotropies. The cluster constraints, given the three different priors on the mass bias, are shown by the filled contours at 68 and 95% confidence, while the dashed black contours give the Planck TT,TE,EE+lowP constraints (hereafter Planck primary CMB, Planck Collaboration XIII 2016); the grey shaded regions add BAO to the CMB. The central value of the WtG mass prior lies at the extreme end of the range used in 2013 (i.e., 1−b = 0.7); with its uncertainty range extending even lower, the tension with primary CMB is greatly reduced, as pointed out by von der Linden et al. (2014b). With similar uncertainty but a central value shifted to 1−b = 0.78, the CCCP mass prior results in greater tension with the primary CMB. The lensing mass prior, finally, implies little bias and hence much greater tension.
The red contours present results from a joint analysis of the cluster counts and the Planck lensing power spectrum (Planck Collaboration XV 2016), adopting our external priors on n_{s} and Ω_{b}h^{2} with the mass bias parameter free and α constrained by the Xray prior. It is interesting to note that these constraints are fully independent of those from the primary CMB, but are in good agreement with them, favouring only slightly lower values for σ_{8}.
Table 3 summarizes our cluster cosmology constraints for the base ΛCDM model for the different mass bias priors. We give the marginalized constraints on Ω_{m} and σ_{8}, as well as their combination that is most tightly constrained by the cluster counts. In addition, in the last line we list constraints when replacing the BAO prior by a prior on H_{0} from direct local measurements (Riess et al. 2011): H_{0} = 73.8 ± 2.4 km s^{1} Mpc^{1}.
7.1.2. Joint Planck 2015 primary CMB and cluster constraints
Mass bias required by the primary CMB.
In Fig. 8 we compare the three prior distributions to the mass bias required by the primary CMB. The latter is obtained as the posterior on 1−b from a joint analysis of the MMF3 cluster counts and the CMB with the mass bias as a free parameter. The bestfit value in this case is 1−b = 0.58 ± 0.04, more than 1σ below the central WtG value. Perfect agreement with the primary CMB would imply that clusters are even more massive than the WtG calibration. This figure most clearly quantifies the tension between the Planck cluster counts and primary CMB.
Fig. 8 Comparison of cluster and primary CMB constraints in the base ΛCDM model, expressed in terms of the mass bias, 1−b. The solid black curve shows the distribution of values required to reconcile the counts and primary CMB in ΛCDM; it is found as the posterior on 1−b from a joint analysis of the Planck cluster counts and primary CMB when leaving the mass bias free. The coloured dashed curves show the three prior distributions on the mass bias listed in Table 2. 
Reionization optical depth.
Primary CMB temperature anisotropies also provide a precise measurement of the parameter combination A_{s}e^{− 2τ}, where τ is the optical depth from Thomson scatter after reionization and A_{s} is the power spectrum normalization on large scales (Planck Collaboration XIII 2016). Lowℓ polarization anisotropies break the degeneracy by constraining τ itself, but this measurement is delicate given the low signal amplitude and difficult systematic effects; it is important, however, in the determination of σ_{8}. It is therefore interesting to compare the Planck primary CMB constraints on τ to those from a joint analysis of the cluster counts and primary CMB without the lowℓ polarization data (lowP). Battye et al. (2015), for instance, pointed out that a lower value for τ than suggested by WMAP could reduce the level of tension between CMB and largescale structure.
The comparison is shown in Fig. 9. We see that the Planck TT + SZ constraints are in good agreement with the value from Planck CMB (i.e., TT,TE,EE+lowP), with the preferred value for WtG slightly higher and CMB lensing pushing towards a lower value. The ordering CMB lensing/CCCP/WtG from lower to higher τ posterior values matches the decreasing level of tension with the primary CMB on σ_{8}. These values remain, however, larger than what is required to fully remove the tension in each case. The posterior distributions for the mass bias are 1−b = 0.60 ± 0.042, 1−b = 0.61 ± 0.049, 1−b = 0.66 ± 0.045, respectively, for WtG, CCCP and CMB lensing, all significantly shifted from the corresponding priors of Table 2. Allowing τ to adjust offers only minor improvement in the tension reflected by Fig. 8. Interestingly, the Planck TT posterior shown in Fig. 8 of Planck Collaboration XIII (2016) peaks at significantly higher values, while our Planck TT + SZ constraints are consistent with the result from Planck TT + lensing, an independent constraint on τ without lowP.
Fig. 9 Constraints on the reionization optical depth, τ. The dashed black curve is the constraint from Planck CMB (i.e., TT,TE,EE+lowP), while the three coloured lines are the posterior distribution on τ from a joint analysis of the cluster counts and Planck TT only for the three different mass bias parameters. 
7.2. Model extensions
7.2.1. Curvature
We consider constraints on spatial curvature that can be set by cluster counts. Our cluster counts combined with BBN and BAO for the CCCP mass prior yield Ω_{K} = −0.06 ± 0.06. This is completely independent of the CMB, but consistent with the CMB plus BAO constraint of Ω_{K} = 0.000 ± 0.002.
7.2.2. Dark energy
Constraints on dark energy and modified gravity based on Planck CMB and external data sets are studied in detail in Planck Collaboration XIV (2016). In Fig. 10 we examine constraints on a constant dark energy equationofstate parameter, w. Analysis of the primary CMB alone results in the highly degenerate grey contours. The degeneracy is broken by adding constraints such as BAO (blue contours) or supernovae distances (rosecolored contours), both picking values around w = −1. The SZ counts (twodimensional likelihood with CCCP prior) only marginally break the degeneracy when combined with the CMB, but when combined with BAO they do yield interesting constraints (green contours) that are consistent with the independent constraints from the primary CMB combined with supernovae. We obtain Ω_{m} = 0.314 ± 0.026 and w = −1.01 ± 0.18 for SZ+BAO, Ω_{m} = 0.306 ± 0.013 and w = −1.10 ± 0.06 for CMB+BAO, and Ω_{m} = 0.306 ± 0.015 and w = −1.10 ± 0.05 for CMB+JLA.
Fig. 10 Constraints on a constant dark energy equationofstate parameter, w. Analysis of the primary CMB alone yields the grey contours that are highly degenerate. Adding either BAO or supernovae to the CMB breaks the degeneracy, giving constraints around w = −1. The green contours are constraints from joint analysis of the SZ counts and BAO; although much less constraining they agree with the CMB+JLA combinations and are completely independent. 
Fig. 11 Parameter constraints on the ΛCDM+nonminimal neutrino mass model. For this study, we adopt the CCCP prior on the mass bias (see Table 2) and leave the scaling exponent, α, free. The green and red shaded regions show, respectively, the 68 and 95% confidence regions for joint analyses of the cluster counts using the primary CMB, and the primary CMB plus the lensing power spectrum. The solid and dashed black contours add to these two cases constraints from BAO. 
7.2.3. ∑ m_{ν}
An important, wellmotivated extension to the base ΛCDM model that clusters can help constrain is a nonminimal sum of neutrino masses, ∑ m_{ν}> 0.06 eV. Given the primary CMB anisotropies, the amplitude of the density perturbations today, characterized by the equivalent linear theory extrapolation, σ_{8}, is model dependent; it is a derived parameter, depending, for example, on the composition of the matter content of the Universe. Cluster abundance, on the other hand, provides a direct measurement of σ_{8} at low redshifts, and comparison to the value derived from the CMB tests the adopted cosmological model.
By freestreaming, neutrinos damp the growth of matter perturbations. Our discussion thus far has assumed the minimum mass for the three known neutrino species. Increasing their mass, ∑ m_{ν}> 0.06 eV, lowers σ_{8} because the neutrinos have larger gravitational influence on the total matter perturbations. This goes in the direction of reconciling tension – the strength of which depends on the mass bias – between the cluster and primary CMB constraints. Cluster abundance, or any measure of σ_{8} at low redshift, is therefore an important cosmological constraint to be combined with those from the primary CMB.
Figure 11 presents a joint analysis of the cluster counts for the CCCP mass bias prior with primary CMB, the Planck lensing power spectrum, and BAO. The results without BAO (green and red shaded contours) allow relatively large neutrino masses, up to ∑ m_{ν} ≈ 0.5 eV; and when adding the lensing power spectrum, a small, broad peak appears in the posterior distribution just above ∑ m_{ν} = 0.2 eV. We also notice some interesting correlations: the amplitude, σ_{8}, anticorrelates with neutrino mass, as does the Hubble parameter, and larger values of α correspond to larger neutrino mass, lower H_{0}, and lower σ_{8}.
As discussed in detail in Planck Collaboration XIII (2016), the anticorrelation with the Hubble parameter maintains the observed acoustic peak scale in the primary CMB. Increasing neutrino mass to simultaneously accommodate the cluster and primary CMB constraints by lowering σ_{8}, while allowed in this joint analysis, would therefore necessarily increase tension with most direct measurements of H_{0} (see discussion in Planck Collaboration XIII 2016). Including the BAO data greatly restricts this possibility, as shown by the solid and dashed black curves.
The solid and dashed, red and black curves in Fig. 12 reproduce the marginalized posterior distributions on ∑ m_{ν} from Fig. 11. The solid blue curve is the result of a similar analysis (CMB+SZ) where, in addition, the artificial parameter A_{L} is allowed to vary. This parameter characterizes the amount of lensing in the temperature power spectrum relative to the best fit model (Planck Collaboration XIII 2016). Planck TT+lowP alone constrains which is in mild tension with the value predicted for the ΛCDM model, A_{L} = 1. In the base ΛCDM model, this parameter is fixed to unity, but it is important to note that it is degenerate with ∑ m_{ν}. Left free, it allows less lensing power, which is also in line with the direct measurement of the lensing power spectrum (labelled as “Lensing PS”) from the fourpoint function (see Planck Collaboration XIII 2016). In that light, we see that adding A_{L} as a free parameter accentuates the peak in the CMB+SZ+Lensing PS posterior. The small internal tension between CMB+SZ and CMB+SZ+A_{L} posteriors may point towards a need for an extension of the minimal sixparameter ΛCDM.
These posteriors lead to the following constraints: We may compare these with the constraints from the primary CMB presented in Planck Collaboration XIII (2016). The Planck primary CMB by itself places an upper limit of ∑ m_{ν}< 0.49 eV (95%), and the addition of BAO tightens this to ∑ m_{ν}< 0.17 eV (95%). Addition of the Planck lensing power spectrum to the primary CMB weakens the constraint to ∑ m_{ν}< 0.59 eV (95%), as we would expect given the results and discussion above. The final constraint adopted by Planck Collaboration XIII (2016), for its robustness to possible remaining low level systematics in the polarization data, is ∑ m_{ν}< 0.23 eV (95%), not too different from the peak suggested in CMB+SZ+lensing PS posterior.
Adding neutrino mass should lower σ_{8}, letting it move towards values favoured by the cluster counts. We might expect that the CMB+SZ combination would therefore favour nonminimal neutrino mass. In spite of this, the green curve only places an upper limit on ∑ m_{ν}. We may understand this by looking at the posterior on the mass bias 1−b in Fig. 11 The allowed values are well separated from the prior distribution (CCCP), meaning that the primary CMB has sufficient statistical weight to strongly override the prior. The lensing power spectrum, in favouring slightly lower σ_{8}, reinforces the cluster trend, so that a peak appears in the posterior for ∑ m_{ν} in the red curve; it is not enough, however, to bring the posterior on the mass bias in line with the prior. This indicates that the tension between the cluster and primary CMB constraints is not fully resolved.
One may then ask, how tight must the prior on the mass bias be to make a difference? To address this question, we performed an analysis assuming a projected tighter prior constraint on the mass bias. The informal target precision for cluster mass calibration with future large lensing surveys, such as Euclid and the Large Synoptic Survey Telescope, is 1%. We therefore consider the impact of a prior of 1−b = 0.78 ± 0.01 on the present Planck cluster cosmology sample in Figs. 12 and 13.
The latter figure compares the constraints from cluster counts for this mass bias to the present primary CMB constraints in the (Ω_{m},σ_{8})plane for the base ΛCDM model. The bold, green dotted curve in Fig. 12 shows the predicted posterior on the neutrino mass from a joint analysis of the present Planck cluster counts and primary CMB with this projected mass bias prior. The same prior on a much larger catalogue would demonstrate a corresponding increase in sensitivity to neutrino mass. This simple projection highlights the importance and value of the more precise cluster cosmology expected in the future, and it provides clear motivation for further significant effort in mass calibration. This effort will continue with larger samples of clusters with gravitational shear measurements, and also with the new technique of CMB lensing cluster mass measurements.
In general, the mass bias is expected to depend on both mass and redshift. A more precise mass calibration study based on much larger samples, e.g., Euclid or LSST, would be required to determine functional forms for the mass bias. A dependence on mass/redshift can be accounted for by appropriate values for α/β in our formalism.
Fig. 12 Constraints on ∑ m_{ν} from a joint analysis of the cluster counts and primary CMB. The solid and dashed, red and black lines reproduce the marginalized posterior distributions from Fig. 11. The solid blue line is the posterior from a similar analysis, but marginalized over the additional parameter A_{L} (see text). If applied to the present Planck cluster cosmology sample, a future mass calibration of 1−b = 0.78 ± 0.01 would result in the bold, dotted green posterior curve. 
Fig. 13 Prediction of cluster constraints with a possible future mass bias prior of 1−b = 0.78 ± 0.01. The black shaded region and dashed contours reproduce the current primary CMB and primary CMB+BAO constraints from Planck for the base ΛCDM model. The red shaded contours present the constraints expected from this mass bias prior applied to the present Planck cluster cosmology sample with the SZ+BAO+BBN data set. 
8. Summary and discussion
Our 2015 analysis incorporates a number of improvements and new information relative to our first study in Planck Collaboration XX (2014). With more data, we have a larger cluster cosmology sample, increased by more than a factor of 2, and we implement a twodimensional likelihood over the counts in both redshift and signaltonoise. We have also performed new tests of the selection function using MCXC and SPT cluster catalogues as truth tables. The selection function from these external checks and internal simulations of the Planck catalogue construction agree with each other and can be reasonably modelled by a simple analytical expression derived by assuming noise is the dominant factor (see Appendix A). One possible systematic effect that warrants further study is IR emission from cluster member galaxies. Finally, we have examined the implications of three recent determinations of the cluster mass bias parameter, 1−b. The twodimensional likelihood with the 2015 catalogue and mass bias priors will be implemented in CosmoMC.
Our analysis confirms the results of the 2013 study. The counts are consistent with those of 2013, illustrated by the agreement in the constraints on Ω_{m} and σ_{8} when using the same SZ observablemass relations (see Fig. 3). The gain in statistical precision is less than expected from the larger catalogue, which is likely related to the fact that the fit to the redshift distribution with the Xray prior on α is only marginal. Our new twodimensional approach yields consistent but more robust constraints than the onedimensional likelihood over just the redshift distribution; it is less sensitive to the slope of the scaling relation, α, and it provides a better fit to the counts than in the onedimensional case.
Using the twodimensional likelihood as our baseline, we extracted new cosmological constraints using three different cluster mass scales represented by the mass bias prior distributions given in Table 2. The first two come from galaxy shear observations of samples of Planck clusters. They differ by about 1σ, with the WtG result favouring larger mass bias. We have also implemented a novel method for measuring cluster masses based on lensing of the CMB temperature anisotropies behind clusters (Melin & Bartlett 2015); it gives a mass bias averaged over the entire cluster cosmology sample, although with larger statistical uncertainty.
As a new method requiring further exploration, we consider CMB lensing less robust at present than galaxy lensing mass measurements, but highly promising. Similar CMBbased mass measurements have recently been published by SPT (Baxter et al. 2015) and ACT (Madhavacheril et al. 2015). The approach is appealing because it is subject to different systematic effects than gravitational shear and because it can be applied to large cluster samples, thanks to the extensive sky coverage of the CMB experiments, with Planck of course covering the entire sky. Gravitational shear surveys will attain large sky coverage in the near future with the Dark Energy Survey (DES), and in the more distant future with the Euclid and WFIRST space missions and the Large Synoptic Survey Telescope.
Our central result from analysis of the 2015 Planck cluster counts is shown in Fig. 7. Depending on the mass bias prior, we find varying degrees of tension with the primary CMB results, as in 2013. The mass bias required to bring the cluster counts and CMB into full agreement is larger than indicated by any of the three priors and corresponds to 1−b = 0.58 ± 0.04. Figure 8 illustrates the situation. The WtG prior almost eliminates the tension, but not quite, while both the CCCP and CMB lensing priors remain in noticeable disagreement. Our largest source of modelling uncertain is, as in 2013, the mass bias.
Tension between low redshift determinations of σ_{8} and the Planck primary CMB results are not unique to the Planck cluster counts. Among SZ cluster surveys, both SPT and ACT are in broad agreement with our findings, the latter depending on which SZmass scaling relation is used, as detailed in our 2013 analysis (Planck Collaboration XX 2014). Furthermore the new SPT cosmological analysis (Bocquet et al. 2015) shows a significant shift between the cluster mass scale determined from the velocity dispersion or Y_{X} and what is needed to satisfy Planck or WMAP9 CMB constraints (e.g., Fig. 2 Bocquet et al. 2015). In a study of the REFLEX Xray luminosity function, Böhringer et al. (2014) also report general agreement with our cluster findings. On the other hand, Mantz et al. (2015) find that their Xray cluster counts, when using the WtG mass calibration, match the primary CMB constraints. Angrick et al. (2015) also find good agreement with the primary CMB constraints, fitting their Xray temperature function with results from hydrodynamical simulations, and Simet et al. (2015) recently measured for the RBC Xray galaxy cluster catalogue.
The situation is thus not yet satisfactory. It is unclear if these modest tensions arise from lowlevel systematics in the astrophysical studies, or are the first glimpse of something more important. The most obvious extension to the base ΛCDM model that could in principle reconcile the differences is a nonminimal sum of neutrino masses. This, unfortunately, does not provide such a straightforward solution. While it is true that adding neutrino mass does lower σ_{8} relative to the base ΛCDM prediction from the primary CMB, it does so at the cost of increasing tension in other parameters; for example, it lowers Planck’s value for the Hubble constant which is already lower than many direct estimates.
Figure 14 highlights these points by showing constraints in the (Ω_{m},σ_{8}) and (H_{0},σ_{8})planes for the CCCP mass bias parameter. Adding variable neutrino mass relaxes constraints from the primary CMB (shaded contours) towards lower σ_{8}, but while simultaneously increasing Ω_{m} and decreasing H_{0}. And the tension remains pronounced, regardless of the neutrino mass.
Another possibility is that baryonic physics influences the latetime evolution of the density perturbations. Strong feedback from active galactic nuclei (van Daalen et al. 2011; Martizzi et al. 2014) can potentially damp growth and lower σ_{8} through expulsion of matter from dark matter halos. This same effect could also reduce the mass of cluster halos and hence the prediction for their abundance, which is based on dark matter only simulations (Cui et al. 2014; Velliscig et al. 2014; Bocquet et al. 2016). It does not appear, however, that these effects are sufficiently large to explain the tension between low redshift and primary CMB constraints hinted at by the different observations (e.g., Cusworth et al. 2014; Costanzi et al. 2014). In addition, the violent feedback necessary for important impact might be difficult to reconcile with observations of the baryon content of dark matter halos (e.g., Planck Collaboration Int. XI 2013; Greco et al. 2015), altough this point is still under discussion (e.g., Planelles et al. 2014).
In conclusion, we return to the main uncertainty in interpretation of the cluster counts, namely the mass bias. It could be argued that the current accuracy is at the level of 10−15%, based on the difference between different analyses and somewhat larger than their quoted statistical uncertainties. Progress will certainly follow with improvement in these measurements. We illustrate the potential impact of a 1% determination of the mass bias in Figs. 12 and 13. Such a result would, depending on the central value, significantly clarify the extent of any tensions and possible necessity for extensions to the base ΛCDM model. This precision is the avowed target of the large lensing surveys, such as Euclid, WFIRST, and LSST. In the shorter term, we may expect valuable movement in this direction from DES and CMB lensing cluster mass measurements.
Fig. 14 Effects of neutrino mass. The open solid magenta contours reproduce our cluster constraints when marginalizing over variable neutrino mass. The violet shaded contours trace the constraints on the base ΛCDM model (with fixed neutrino mass), while the other shaded regions give constraints from the primary CMB only or combined with lensing and BAO when adding and marginalizing over variable neutrino mass. In this figure, the CMB likelihood is based on Planck TT, TE, EE+lowP while only Planck TT+lowP is used in the equivalent figure in Planck Collaboration XIII (2016). 
Planck (http://www.esa.int/Planck) is a project of the European Space Agency (ESA) with instruments provided by two scientific consortia funded by ESA member states and led by Principal Investigators from France and Italy, telescope reflectors provided through a collaboration between ESA and a scientific consortium led and funded by Denmark, and additional contributions from NASA (USA).
This form assumes, as we do throughout, that the distribution depends on z and M_{500} only through the mean value , specifically, that the intrinsic scatter, σ_{lnY}, of Eq. (9) is constant.
In fact, this expectation is violated to a small degree. As discussed in Planck Collaboration XXVII (2016), there appears to be a small population of Xray underluminous clusters.
Acknowledgments
The Planck Collaboration acknowledges the support of: ESA; CNES, and CNRS/INSUIN2P3INP (France); ASI, CNR, and INAF (Italy); NASA and DoE (USA); STFC and UKSA (UK); CSIC, MINECO, JA and RES (Spain); Tekes, AoF, and CSC (Finland); DLR and MPG (Germany); CSA (Canada); DTU Space (Denmark); SER/SSO (Switzerland); RCN (Norway); SFI (Ireland); FCT/MCTES (Portugal); ERC and PRACE (EU). A description of the Planck Collaboration and a list of its members, indicating which technical or scientific activities they have been involved in, can be found at http://www.cosmos.esa.int/web/planck/planckcollaboration.
References
 Anderson, L., Aubourg, E., Bailey, S., et al. 2012, MNRAS, 427, 3435 [Google Scholar]
 Anderson, L., Aubourg, É., Bailey, S., et al. 2014, MNRAS, 441, 24 [NASA ADS] [CrossRef] [Google Scholar]
 Andreon, S. 2014, A&A, 570, L10 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
 Angrick, C., Pace, F., Bartelmann, M., & Roncarelli, M. 2015, MNRAS, 454, 1687 [NASA ADS] [CrossRef] [Google Scholar]
 Angulo, R. E., Springel, V., White, S. D. M., et al. 2012, MNRAS, 426, 2046 [CrossRef] [Google Scholar]
 Arnaud, M., Pratt, G. W., Piffaretti, R., et al. 2010, A&A, 517, A92 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
 Battye, R. A., Charnock, T., & Moss, A. 2015, Phys. Rev. D, 91, 103508 [NASA ADS] [CrossRef] [Google Scholar]
 Baxter, E. J., Keisler, R., Dodelson, S., et al. 2015, ApJ, 806, 247 [NASA ADS] [CrossRef] [Google Scholar]
 Becker, M. R., & Kravtsov, A. V. 2011, ApJ, 740, 25 [NASA ADS] [CrossRef] [Google Scholar]
 Benson, B. A., de Haan, T., Dudley, J. P., et al. 2013, ApJ, 763, 147 [NASA ADS] [CrossRef] [Google Scholar]
 Bersanelli, M., Mandolesi, N., Butler, R. C., et al. 2010, A&A, 520, A4 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
 Betoule, M., Kessler, R., Guy, J., et al. 2014, A&A, 568, A22 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
 Beutler, F., Blake, C., Colless, M., et al. 2011, MNRAS, 416, 3017 [NASA ADS] [CrossRef] [Google Scholar]
 Bleem, L. E., Stalder, B., de Haan, T., et al. 2015, ApJS, 216, 27 [NASA ADS] [CrossRef] [Google Scholar]
 Bocquet, S., Saro, A., Mohr, J. J., et al. 2015, ApJ, 799, 214 [NASA ADS] [CrossRef] [Google Scholar]
 Bocquet, S., Saro, A., Dolag, K., & Mohr, J. J. 2016, MNRAS, 456, 2361 [Google Scholar]
 Böhringer, H., Chon, G., & Collins, C. A. 2014, A&A, 570, A31 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
 Carlstrom, J. E., Ade, P. A. R., Aird, K. A., et al. 2011, PASP, 123, 568 [NASA ADS] [CrossRef] [Google Scholar]
 Carvalho, P., Rocha, G., & Hobson, M. P. 2009, MNRAS, 393, 681 [NASA ADS] [CrossRef] [Google Scholar]
 Chamballu, A., Bartlett, J. G., & Melin, J.B. 2012, A&A, 544, A40 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
 Costanzi, M., Sartoris, B., Viel, M., & Borgani, S. 2014, J. Cosmol. Astropart. Phys. , 10, 81 [NASA ADS] [CrossRef] [Google Scholar]
 Cui, W., Borgani, S., & Murante, G. 2014, MNRAS, 441, 1769 [CrossRef] [Google Scholar]
 Cusworth, S. J., Kay, S. T., Battye, R. A., & Thomas, P. A. 2014, MNRAS, 439, 2485 [NASA ADS] [CrossRef] [Google Scholar]
 Greco, J. P., Hill, J. C., Spergel, D. N., & Battaglia, N. 2015, ApJ, 808, 151 [NASA ADS] [CrossRef] [Google Scholar]
 Hasselfield, M., Hilton, M., Marriage, T. A., et al. 2013, J. Cosmol. Astropart. Phys. , 7, 8 [Google Scholar]
 Hoekstra, H., Herbonnet, R., Muzzin, A., et al. 2015, MNRAS, 449, 685 [NASA ADS] [CrossRef] [Google Scholar]
 Hu, W., & Kravtsov, A. V. 2003, ApJ, 584, 702 [NASA ADS] [CrossRef] [Google Scholar]
 Lamarre, J., Puget, J., Ade, P. A. R., et al. 2010, A&A, 520, A9 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
 Lewis, A., & Challinor, A. 2006, Phys. Rep., 429, 1 [Google Scholar]
 Madhavacheril, M., Sehgal, N., Allison, R., et al. 2015, Phys. Rev. Lett., 114, 151302 [NASA ADS] [CrossRef] [Google Scholar]
 Mantz, A. B., von der Linden, A., Allen, S. W., et al. 2015, MNRAS, 446, 2205 [NASA ADS] [CrossRef] [Google Scholar]
 Marrone, D. P., Smith, G. P., Okabe, N., et al. 2012, ApJ, 754, 119 [NASA ADS] [CrossRef] [Google Scholar]
 Martizzi, D., Mohammed, I., Teyssier, R., & Moore, B. 2014, MNRAS, 440, 2290 [NASA ADS] [CrossRef] [Google Scholar]
 Melin, J.B., & Bartlett, J. G. 2015, A&A, 578, A21 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
 Melin, J., Bartlett, J. G., & Delabrouille, J. 2006, A&A, 459, 341 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
 Meneghetti, M., Rasia, E., Merten, J., et al. 2010, A&A, 514, A93 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
 Mennella, A., Butler, R. C., Curto, A., et al. 2011, A&A, 536, A3 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
 Merten, J., Meneghetti, M., Postman, M., et al. 2015, ApJ, 806, 4 [NASA ADS] [CrossRef] [Google Scholar]
 Nagai, D., Kravtsov, A. V., & Vikhlinin, A. 2007, ApJ, 668, 1 [NASA ADS] [CrossRef] [Google Scholar]
 Navarro, J. F., Frenk, C. S., & White, S. D. M. 1997, ApJ, 490, 493 [NASA ADS] [CrossRef] [Google Scholar]
 Olive, K. A. 2013, in AIP Conf. Ser., 1548, eds. L. A. UrenãLópez, R. BecerrilBárcenas, & R. LinaresRomero, 116 [Google Scholar]
 Padmanabhan, N., Xu, X., Eisenstein, D. J., et al. 2012, MNRAS, 427, 2132 [NASA ADS] [CrossRef] [MathSciNet] [Google Scholar]
 Piffaretti, R., & Valdarnini, R. 2008, A&A, 491, 71 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
 Piffaretti, R., Arnaud, M., Pratt, G. W., Pointecouteau, E., & Melin, J.B. 2011, A&A, 534, A109 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
 Planck Collaboration I. 2011, A&A, 536, A1 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
 Planck Collaboration I. 2014, A&A, 571, A1 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
 Planck Collaboration XVI. 2014, A&A, 571, A16 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
 Planck Collaboration XX. 2014, A&A, 571, A20 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
 Planck Collaboration XXIX. 2014, A&A, 571, A29 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
 Planck Collaboration I. 2016, A&A, 594, A1 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
 Planck Collaboration II. 2016, A&A, 594, A2 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
 Planck Collaboration III. 2016, A&A, 594, A3 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
 Planck Collaboration IV. 2016, A&A, 594, A4 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
 Planck Collaboration V. 2016, A&A, 594, A5 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
 Planck Collaboration VI. 2016, A&A, 594, A6 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
 Planck Collaboration VII. 2016, A&A, 594, A7 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
 Planck Collaboration VIII. 2016, A&A, 594, A8 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
 Planck Collaboration IX. 2016, A&A, 594, A9 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
 Planck Collaboration X. 2016, A&A, 594, A10 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
 Planck Collaboration XI. 2016, A&A, 594, A11 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
 Planck Collaboration XII. 2016, A&A, 594, A12 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
 Planck Collaboration XIII. 2016, A&A, 594, A13 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
 Planck Collaboration XIV. 2016, A&A, 594, A14 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
 Planck Collaboration XV. 2016, A&A, 594, A15 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
 Planck Collaboration XVI. 2016, A&A, 594, A16 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
 Planck Collaboration XVII. 2016, A&A, 594, A17 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
 Planck Collaboration XVIII. 2016, A&A, 594, A18 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
 Planck Collaboration XIX. 2016, A&A, 594, A19 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
 Planck Collaboration XX. 2016, A&A, 594, A20 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
 Planck Collaboration XXI. 2016, A&A, 594, A21 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
 Planck Collaboration XXII. 2016, A&A, 594, A22 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
 Planck Collaboration XXIII. 2016, A&A, 594, A23 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
 Planck Collaboration XXIV. 2016, A&A, 594, A24 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
 Planck Collaboration XXV. 2016, A&A, 594, A25 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
 Planck Collaboration XXVI. 2016, A&A, 594, A26 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
 Planck Collaboration XXVII. 2016, A&A, 594, A27 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
 Planck Collaboration XXVIII. 2016, A&A, 594, A28 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
 Planck Collaboration Int. III. 2013, A&A, 550, A129 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
 Planck Collaboration Int. XI. 2013, A&A, 557, A52 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
 Planck HFI Core Team 2011, A&A, 536, A4 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
 Planelles, S., Borgani, S., Fabjan, D., et al. 2014, MNRAS, 438, 195 [NASA ADS] [CrossRef] [Google Scholar]
 Postman, M., Coe, D., Benítez, N., et al. 2012, ApJS, 199, 25 [NASA ADS] [CrossRef] [Google Scholar]
 Rasia, E., Ettori, S., Moscardini, L., et al. 2006, MNRAS, 369, 2013 [NASA ADS] [CrossRef] [Google Scholar]
 Rasia, E., Meneghetti, M., Martino, R., et al. 2012, New J. Phys., 14, 055018 [NASA ADS] [CrossRef] [Google Scholar]
 Rasia, E., Lau, E. T., Borgani, S., et al. 2014, ApJ, 791, 96 [NASA ADS] [CrossRef] [Google Scholar]
 Reichardt, C. L., Stalder, B., Bleem, L. E., et al. 2013, ApJ, 763, 127 [NASA ADS] [CrossRef] [Google Scholar]
 Riess, A. G., Macri, L., Casertano, S., et al. 2011, ApJ, 730, 119 [NASA ADS] [CrossRef] [Google Scholar]
 Rozo, E., Bartlett, J. G., Evrard, A. E., & Rykoff, E. S. 2014a, MNRAS, 438, 78 [NASA ADS] [CrossRef] [Google Scholar]
 Rozo, E., Evrard, A. E., Rykoff, E. S., & Bartlett, J. G. 2014b, MNRAS, 438, 62 [NASA ADS] [CrossRef] [Google Scholar]
 Rozo, E., Rykoff, E. S., Bartlett, J. G., & Evrard, A. 2014c, MNRAS, 438, 49 [NASA ADS] [CrossRef] [Google Scholar]
 Simet, M., Battaglia, N., Mandelbaum, R., & Seljak, U. 2015, ArXiv eprints [arXiv:1502.01024] [Google Scholar]
 Steigman, G. 2008, ArXiv eprints [arXiv:0807.3004] [Google Scholar]
 Swetz, D. S., Ade, P. A. R., Amiri, M., et al. 2011, ApJS, 194, 41 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
 Tauber, J. A., Mandolesi, N., Puget, J., et al. 2010, A&A, 520, A1 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
 Tinker, J., Kravtsov, A. V., Klypin, A., et al. 2008, ApJ, 688, 709 [NASA ADS] [CrossRef] [Google Scholar]
 Umetsu, K., Medezinski, E., Nonino, M., et al. 2014, ApJ, 795, 163 [NASA ADS] [CrossRef] [Google Scholar]
 van Daalen, M. P., Schaye, J., Booth, C. M., & Dalla Vecchia, C. 2011, MNRAS, 415, 3649 [NASA ADS] [CrossRef] [Google Scholar]
 Velliscig, M., van Daalen, M. P., Schaye, J., et al. 2014, MNRAS, 442, 2641 [NASA ADS] [CrossRef] [Google Scholar]
 von der Linden, A., Allen, M. T., Applegate, D. E., et al. 2014a, MNRAS, 439, 2 [NASA ADS] [CrossRef] [Google Scholar]
 von der Linden, A., Mantz, A., Allen, S. W., et al. 2014b, MNRAS, 443, 1973 [NASA ADS] [CrossRef] [Google Scholar]
 Watson, W. A., Iliev, I. T., D’Aloisio, A., et al. 2013a, MNRAS, 433, 1230 [Google Scholar]
 Watson, W. A., Iliev, I. T., D’Aloisio, A., et al. 2013b, MNRAS, submitted [arXiv:1212.0095v1] [Google Scholar]
 Zaldarriaga, M., & Seljak, U. 1999, Phys. Rev. D, 59, 123507 [NASA ADS] [CrossRef] [Google Scholar]
Appendix A: Modelling uncertainties
We examine the robustness of our cosmological constraints to modelling uncertainties. We first consider sensitivity to the cosmological sample and to our modelling of the completeness function in Sect. A.1, and then look at the effect of using a different mass function in Sect. A.2. In Sect. A.3, we show that our constraints are robust against redshift evolution of the scaling relations.
Appendix A.1: Choice of the sample and selection function
For our baseline analysis, we use the MMF3 cosmological sample and its associated completeness based on the analytical approximation using the error function (Eq. (14)). In Fig. A.1 we show how the Monte Carlo determined selection function changes the cosmological constraints (labelled QA for “Quality Assessment” in the figure). We also present the constraints obtained from the intersection sample defined in Sect. 2. The figure is based on the 1D N(z) likelihood, for which the Monte Carlo completeness can be easily computed, and we use the baseline SZ+BAO+BBN data set and fix 1−b = 0.8. The MMF3 ERF contour is thus close to the q = 6 contour of Fig. 3, the only difference being that σ_{lnY} is fixed to zero in Fig. A.1. while it is constrained by the Table 1 prior in Fig. 3. The impact of adopting the intersection sample and/or the QA completeness function is small (< 0.5σ) for both Ω_{m} and σ_{8}.
Fig. A.1 Robustness to the choice of cluster sample and selection function model. The shaded contours give the cosmological constraints from the 2015 MMF3 cluster catalogue using the analytical (error function, ERF) selection function model (grey), the MMF3 Monte Carlo selection function (red), and the Monte Carlo selection function for the intersection sample (blue). Our final constraints are obtained from the MMF3 ERF model. For this comparison, we adopt the SZ+BAO+BBN data set and we fix 1−b = 0.8. 
Appendix A.2: Mass function
We use the Tinker et al. (2008) mass function for our baseline analysis. To characterize the influence of this choice, we examine constraints when adopting the mass function from Watson et al. (2013a) instead. We employ our twodimensional N(z,q) likelihood (with the CCCP mass bias prior and α constrained) and combine with BAO and BBN prior constraints, showing the result in Fig. A.2. The Tinker et al. contour of Fig. A.2 is thus identical to the N(z,q) contour with α constrained, as given in Fig. 5. The new mass function shifts our constraints by about 1σ towards higher Ω_{m} and lower σ_{8}, along the main degeneracy line, hence increasing the tension with the Planck primary CMB constraints. We note that we use the general fit from Eq. (12) of Watson et al. (2013a) (independent of redshift). This was not the case for our 2013 paper (Planck Collaboration XX 2014) where we adopted the AHF fit with parameters varying with redshift in the first posted version of the paper (see Watson et al. 2013b, on ArXiv.org), which was subsequently found to be incorrect.
Fig. A.2 Robustness to the choice of mass function. The grey shaded contours give the cosmological constraints when using the Tinker et al. (2008) mass function, corresponding to our final result. This is compared to constraints obtained when using the Watson et al. (2013a) mass function, shown as the red shaded contours. In this figure we adopt the SZ+BAO+BBN data set and the CCCP mass bias prior. 
Appendix A.3: Redshift evolution of the YM relation
Throughout our baseline analysis, we fix the redshift evolution exponent β = 0.66 (selfsimilar prediction) in Eq. (7). Here we examine the impact of allowing this parameter to vary. Constraints when leaving β free are shown in Fig. A.3. The “β fixed, α constrained” case corresponds to the twodimensional N(z,q) likelihood (CCCP mass bias prior and α constrained) combined with BAO and BBN, as in Fig. A.2. This contour is also identical to the N(z,q) contour with α constrained, shown in Fig. 5. For the “β constrained” cases, β is allowed to vary over the range 0.66 ± 0.50 (Table 1). This increases the size of our constraints along the major degeneracy between Ω_{m} and σ_{8}, but does not bring them into any closer agreement with the primary CMB.
Andreon (2014) recently reanalyzed the subsample of 71 clusters used in Planck Collaboration XX (2014). Through a joint fit to the normalization and redshift evolution of the Y−M relation, he found a significant detection of nonstandard redshift evolution. It is possible that this conclusion is driven by systematic effects in the Xray and SZ measurements. In particular, lowz objects have larger angular sizes, and so measurement of their Xray and SZ quantities are subject to different systematic effects compared to equivalent measurements for highz objects. Appendix D of Planck Collaboration XXIX (2014) includes a comprehensive discussion of the various systematic effects that may have an impact on measurement of the Y−M relation. Given these caveats, we believe that it is premature to draw definitive conclusions on the evolution of the Y−M relation.
Fig. A.3 Robustness to redshift evolution in the SZmass scaling relation. The different contours show the constraints when relaxing the redshift evolution exponent, β, of Eq. (7). The black contours result from fixing β = 0.66, our fiducial value throughout, with α constrained by the Gaussian Xray prior of Table 1. Applying a Gaussian the prior on β instead, from Table 1, produces the blue contours, while the red contours result when we also leave α free. In this figure we adopt the SZ+BAO+BBN data set and the CCCP mass bias prior. 
Appendix A.4: Slope of the Y−M relation
The onedimensional analysis of the redshift distribution in Sect. 6.2 preferred steeper values of the Y−M scaling exponent, α, than indicated by Xray studies of local clusters (see Fig. 5, case α free, black contours and curves). This seems to be related to the flattening in the redshift distribution around z = 0.2 seen in Figs. 4 and 6. To explore this, we separately extracted parameter constraints from low (z< 0.2) and high (z> 0.2) redshift bin sets (two and eight bins, respectively) when leaving α free.
Figure A.4 compares the constraints from these splits to those from the full bin set (with α free). Neither the low nor
the high redshift bin set on its own prefers the steep slope – the posteriors peak near the Xray value, i.e, the clusters in each subsample follow the expected scaling relation. Cosmological constraints from the lowz data broaden compared to the full data set, and bimodality appears in the distribution for Ω_{m}. The highz data, on the other hand, loose precision on α, although the peak in the posterior shifts to the Xray value, but maintain constraints on the cosmological parameters that are similar to those from the full data set.
Only the combined data covering the full redshift range prefer the steep slope. The parameters are coupled in a subtle interplay that favours high values of α and moves the cosmological parameters along the degeneracy ridges to settle on their highz values. These degeneracies mask the interplay and unfortunately make further exploration difficult with the present data set.
Fig. A.4 Comparison of constraints from the full redshift distribution (in black) to those from the redshift split: two bins at z< 0.2 (in purple) and eight bins at z> 0.2 (in blue). The constraints are obtained with the onedimensional likelihood when leaving α free, and the black curves and contours reproduce those from Fig. 5. 
All Tables
All Figures
Fig. 1 Massredshift distribution of the Planck cosmological samples colourcoded by their signaltonoise, q. The baseline MMF3 2015 cosmological sample is shown as the small filled circles. Objects which were in the MMF3 2013 cosmological sample are marked by crosses, while those in the 2015 intersection sample are shown as open circles. The final samples are defined by q> 6. The mass M_{Yz} is the Planck mass proxy (see text; Arnaud et al., in prep.). 

In the text 
Fig. 2 Cluster mass scale determined by CMB lensing. We show the ratio of cluster lensing mass, M_{lens}, to the SZ mass proxy, M_{Yz}, as a function of the mass proxy for clusters in the MMF3 2015 cosmology sample. The cluster mass is measured through lensing of CMB temperature anisotropies in the Planck data (Melin & Bartlett 2015). Individual mass measurements have low signaltonoise, but we determine a mean ratio for the sample of M_{lens}/M_{Yz} = 1 / (1−b) = 0.99 ± 0.19. For clarity, only some of the error bars are plotted (see text). 

In the text 
Fig. 3 Contours at 95% for different signaltonoise thresholds, q = 8.5, 7, and 6, applied to the 2015 MMF3 cosmology sample for the SZ+BAO+BBN data set. The contours are compatible with the 2013 constraints (Planck Collaboration XX 2014), shown as the filled, light grey ellipses at 68 and 95% (for the BAO and BBN priors of Sect 5.2; see text). The 2015 catalogue thresholded at q> 8.5 has a similar number of clusters (190) as the 2013 catalogue (189). This comparison is made using the analytical errorfunction model for completeness and adopts the reference observablemass scaling relation of the 2013 analysis (1−b = 0.8, see text). The redshift distributions of the bestfit models are shown in Fig. 4. For this figure and Fig. 4, we use the onedimensional likelihood over the redshift distribution, dN/ dz (Eq. (4)). 

In the text 
Fig. 4 Comparison of observed counts (points with error bars) with predictions of the bestfit models (solid lines) from the onedimensional likelihood for three different thresholds applied to the 2015 MMF3 cosmology sample. The mismatch between observed and predicted counts in the second and third lowest redshift bins, already noticed in the 2013 analysis, increases at lower thresholds, q. The bestfit models are defined by the constraints shown in Fig. 3. For this figure and Fig. 3, we use our onedimensional likelihood over the redshift distribution, dN/ dz (Eq. (4)), with the mass biased fixed at (1−b) = 0.8. 

In the text 
Fig. 5 Comparison of constraints from the onedimensional (dN/ dz) and twodimensional (dN/ dzdq) likelihoods on cosmological parameters and the scaling relation mass exponent, α. This comparison uses the MMF3 catalogue, the CCCP prior on the mass bias and the SZ+BAO+BBN data set. The corresponding bestfit model redshift distributions are shown in Fig. 6. 

In the text 
Fig. 6 Redshift distribution of bestfit models from the four analysis cases shown in Fig. 5. The observed counts in the MMF3 catalogue (q> 6) are plotted as the red points with error bars, and as in Fig. 5 we adopt the CCCP mass prior with the SZ+BAO+BBN data set. 

In the text 
Fig. 7 Comparison of constraints from the CMB to those from the cluster counts in the (Ω_{m},σ_{8})plane. The green, blue and violet contours give the cluster constraints (twodimensional likelihood) at 68 and 95% for the WtG, CCCP, and CMB lensing mass calibrations, respectively, as listed in Table 2. These constraints are obtained from the MMF3 catalogue with the SZ+BAO+BBN data set and α free (hence the SZα notation). Constraints from the Planck TT, TE, EE+lowP CMB likelihood (hereafter, Planck primary CMB) are shown as the dashed contours enclosing 68 and 95% confidence regions (Planck Collaboration XIII 2016), while the grey shaded region also includes BAO. The red contours give results from a joint analysis of the cluster counts and the Planck lensing power spectrum (Planck Collaboration XV 2016), adopting our external priors on n_{s} and Ω_{b}h^{2} with the mass bias parameter free and α constrained by the Xray prior (hence the SZ notation without the subscript α). 

In the text 
Fig. 8 Comparison of cluster and primary CMB constraints in the base ΛCDM model, expressed in terms of the mass bias, 1−b. The solid black curve shows the distribution of values required to reconcile the counts and primary CMB in ΛCDM; it is found as the posterior on 1−b from a joint analysis of the Planck cluster counts and primary CMB when leaving the mass bias free. The coloured dashed curves show the three prior distributions on the mass bias listed in Table 2. 

In the text 
Fig. 9 Constraints on the reionization optical depth, τ. The dashed black curve is the constraint from Planck CMB (i.e., TT,TE,EE+lowP), while the three coloured lines are the posterior distribution on τ from a joint analysis of the cluster counts and Planck TT only for the three different mass bias parameters. 

In the text 
Fig. 10 Constraints on a constant dark energy equationofstate parameter, w. Analysis of the primary CMB alone yields the grey contours that are highly degenerate. Adding either BAO or supernovae to the CMB breaks the degeneracy, giving constraints around w = −1. The green contours are constraints from joint analysis of the SZ counts and BAO; although much less constraining they agree with the CMB+JLA combinations and are completely independent. 

In the text 
Fig. 11 Parameter constraints on the ΛCDM+nonminimal neutrino mass model. For this study, we adopt the CCCP prior on the mass bias (see Table 2) and leave the scaling exponent, α, free. The green and red shaded regions show, respectively, the 68 and 95% confidence regions for joint analyses of the cluster counts using the primary CMB, and the primary CMB plus the lensing power spectrum. The solid and dashed black contours add to these two cases constraints from BAO. 

In the text 
Fig. 12 Constraints on ∑ m_{ν} from a joint analysis of the cluster counts and primary CMB. The solid and dashed, red and black lines reproduce the marginalized posterior distributions from Fig. 11. The solid blue line is the posterior from a similar analysis, but marginalized over the additional parameter A_{L} (see text). If applied to the present Planck cluster cosmology sample, a future mass calibration of 1−b = 0.78 ± 0.01 would result in the bold, dotted green posterior curve. 

In the text 
Fig. 13 Prediction of cluster constraints with a possible future mass bias prior of 1−b = 0.78 ± 0.01. The black shaded region and dashed contours reproduce the current primary CMB and primary CMB+BAO constraints from Planck for the base ΛCDM model. The red shaded contours present the constraints expected from this mass bias prior applied to the present Planck cluster cosmology sample with the SZ+BAO+BBN data set. 

In the text 
Fig. 14 Effects of neutrino mass. The open solid magenta contours reproduce our cluster constraints when marginalizing over variable neutrino mass. The violet shaded contours trace the constraints on the base ΛCDM model (with fixed neutrino mass), while the other shaded regions give constraints from the primary CMB only or combined with lensing and BAO when adding and marginalizing over variable neutrino mass. In this figure, the CMB likelihood is based on Planck TT, TE, EE+lowP while only Planck TT+lowP is used in the equivalent figure in Planck Collaboration XIII (2016). 

In the text 
Fig. A.1 Robustness to the choice of cluster sample and selection function model. The shaded contours give the cosmological constraints from the 2015 MMF3 cluster catalogue using the analytical (error function, ERF) selection function model (grey), the MMF3 Monte Carlo selection function (red), and the Monte Carlo selection function for the intersection sample (blue). Our final constraints are obtained from the MMF3 ERF model. For this comparison, we adopt the SZ+BAO+BBN data set and we fix 1−b = 0.8. 

In the text 
Fig. A.2 Robustness to the choice of mass function. The grey shaded contours give the cosmological constraints when using the Tinker et al. (2008) mass function, corresponding to our final result. This is compared to constraints obtained when using the Watson et al. (2013a) mass function, shown as the red shaded contours. In this figure we adopt the SZ+BAO+BBN data set and the CCCP mass bias prior. 

In the text 
Fig. A.3 Robustness to redshift evolution in the SZmass scaling relation. The different contours show the constraints when relaxing the redshift evolution exponent, β, of Eq. (7). The black contours result from fixing β = 0.66, our fiducial value throughout, with α constrained by the Gaussian Xray prior of Table 1. Applying a Gaussian the prior on β instead, from Table 1, produces the blue contours, while the red contours result when we also leave α free. In this figure we adopt the SZ+BAO+BBN data set and the CCCP mass bias prior. 

In the text 
Fig. A.4 Comparison of constraints from the full redshift distribution (in black) to those from the redshift split: two bins at z< 0.2 (in purple) and eight bins at z> 0.2 (in blue). The constraints are obtained with the onedimensional likelihood when leaving α free, and the black curves and contours reproduce those from Fig. 5. 

In the text 
Current usage metrics show cumulative count of Article Views (fulltext article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.
Data correspond to usage on the plateform after 2015. The current usage metrics is available 4896 hours after online publication and is updated daily on week days.
Initial download of the metrics may take a while.