Mass and selection biases of galaxy clusters: a multi-probe approach

Galaxy clusters are fascinating astrophysical objects with a great interest for cosmological applications.

Baryons, detectable at different wavelengths, represent only small amount in the mass budget (12% in the diffuse IntraCluster Medium, ICM, and 3% confined in galaxies), while Dark Matter, DM, is the dominant ingredient (85%) but unfortunately only traceable through its gravitational impact. 

Up to now, clusters have been mostly selected by their minority component: directly through the emission from galaxies or the hot ICM, or through the scattering of the Cosmic Microwave Background (CMB) on the ICM (Sunayev-Zeldovich, SZ, effect). These selection criteria are convenient, have provided us with large samples at various redshifts and have helped us establish many of the cluster properties. However, the samples are biased and the amplitude of the bias is yet to be determined. 

Furthermore, knowledge of the cluster mass is needed for both cosmological and astrophysical studies, but mass cannot be directly observed and needs to be derived from the collected photons. 

Our understanding of galaxy cluster physics, and our ability to use clusters to constrain cosmological parameters, are limited by mass and selection biases of the studied samples. 

Given the critical role played by the mass and selection biases, in this project we (Unit 1) exploit the only sample available so far that contains a sizable fraction of low surface brightness clusters missed in X-ray- and SZ-selected surveys. We aim at better determining the role of the selection bias (against low surface brightness clusters) in scaling relations of cosmological and astrophysical interest with the assistance of our data and hydrodynamic simulations; and we (Unit 2) investigate the amount of mass bias by using multi-probe approaches based on high-resolution and high-sensitivity observations with the support of state-of-the-art hydrodynamical simulations.

With this project, we can address, uniquely and for the first time, the mass bias of a sample non-ICM biased and the impact of the ICM selection on the mass bias by comparing ICM-biased and ICM-unbiased samples. We hope to overcome the limitation of literature studies that address mass bias and selection effects independently.

Projects related on Mass inference and possible biases - Research Unit 1

Almost all the following projects benefit from the on-going active collaborations: The300, NIKA2, CHEX-MATE and EURA NOVA.

• Cluster mass inference could be impacted by cluster dynamical state. Studying the most efficient morphological proxies, or a combination of them, to infer the dynamical state with survey-quality data and with hydrodynamical simulations is mandatory, see Cialone G. et al. 2018, De Luca F. et al. 2021, Zhang B. et al. 2022.

A new and promising approach based on Zernike Polynomials (ZP), already validated on SZ maps generated by synthetic clusters of The300 in Capalbo V. et al. 2021, is extended to different resolutions and wavelengths images and real data to infer the dynamical state.

• 1. 1. 1. Planck nearby clusters (109 clusters at z<0.1), Capalbo V. et al. submitted to A&A;

        2. NIKA2 Large Program Sunyaev Zeldovich dataset (35 clusters at 0.5<z<0.9 observed at 150 & 260 GHz), Pappalardo E. et al. in preparation;

        3. CHEX-MATE sample (118 clusters  observed by XMM-Newton), Benincasa A. et al. in preparation;

        4. optical images (galaxy number density maps), Ferragamo A. et al. on-going analysis.

Other different metrics can be used to classify cluster dynamical state: a possibility has been the recent use of two different statistical approaches –Principal Component Analysis (PCA) and Uniform Manifold Approximation and Projection (UMAP), see Haggar R. et al. 2024.

• Inference of a minimally biased total mass by applying Machine Learning techniques. This work is based on a strong/old collaboration with the Universidad Autonoma de Madrid (G. Yepes, W. Cui, D. de Andres and collabs) and EURA NOVA, a company working on big data. We started with an interesting work to infer the total mass of Planck clusters, i.e. without any mass bias, by training a CNN model on The300 Compton-y maps and then by applying it on real Planck maps, see de Andres D. et al. 2022. We then explored the reconstruction of unbiased 3D total (and gas) mass radial profiles starting from mock Compton-y maps, see Ferragamo A. et al. 2023. Now we are about to take few steps forward with the following projects: 

  1. 1. 1. Generating observations of galaxy clusters from Dark-Matter-only simulations by Deep Learning, Caro A. et al. 2024 submitted to RASti;

        2. Reconstruction of total mass radial profiles of NIKA2 Twin Samples by ML model, Ferragamo A. et al. on-going analysis;

        3. How much unbiased Planck clusters masses, inferred by CNN model, could impact cosmological parameters? Wicker R. et al. og-going analysis.

• Impact of filaments on cluster properties, such as mass and bias. How much the filamentary structures around clusters are correlated with their properties? 

  1. 1. 1. Cluster connectivity, estimated by gas density field, is recostructured in The300 regions, Santoni S. et al. 2024 submitted to A&A.

        2. Mapping gas filaments by SZ and X-ray signals in clusters outskirts, Santoni S. et al. on-going analysis.

• Estimate of cluster mass by NIKA2 observations. Among the several projects on-going within the NIKA2 LPSZ team, we are investigating:

  1. 1. 1. the IntraCluster Medium and dynamical state of ACT-CL J0240.0+0116 using a multi-wavelength approach, Paliwal A. et al. 2024 submitted to A&A;

        2. the gas pressure profiles in the NIKA2 redshift range (0.5-0.9) reconstructed from NIKA2 Twin Sample mock images at 150 GHz: comparison with catalog profiles and available models in literature, Paliwal A. et al. 2024 on-going analysis.

• The Sparsity in clusters, the ratio of spherical halo masses estimated at radii enclosing different overdensities, is useful to infer constraints on cosmological model parameters but it's important to take care of the impact of possible mass bias, such as the hydrostatic one, see Corasaniti S. et al. 2024 under submission.

  1. 1. 1. To investigate the correlation between Sparsity, mass bias and cluster dynamical state., TBD in the planning stage.

• For very distant objects, where X-ray spectromets are starved of photons, it is difficult to obtain an accurate map of the temperature of the gas in the clusters. The combination of SZ and X-ray intensity observations will allow to reconstruct such kind of maps.

  1. 1. 1. We are validating a model to recover mass-weighted temperature maps using simulated SZ and X-ray images of The300 clusters affected by observational impacts of NIKA2 and XMM-Newton instruments. Wicker R. et al. on-going analysis. 

        2. The application on NIKA2 LPSZ with XMM-Newton maps is just around the corner ...