Convolutional neural network-reconstructed velocity for kinetic SZ detection

¹ Université Paris-Saclay, CNRS, Institut d’Astrophysique Spatiale, Bâtiment 121 91405, Orsay, France
e-mail: hideki.tanimura@ias.u-psud.fr
² Instituto de Astrofísica de Canarias, 38205 Tenerife, Spain
³ University of La Laguna, 38206 Tenerife, Spain
⁴ ARCO (Astrophysics Research Center), Department of Natural Sciences, The Open University of Israel, 1 University Road, PO Box 808, Ra’anana 4353701, Israel
⁵ Kapteyn Astronomical Institute, University of Groningen, PO Box 800, 9700 AV Groningen, The Netherlands

Received: 5 January 2022
Accepted: 5 April 2022

Abstract

We report the detection of the kinetic Sunyaev-Zel’dovich (kSZ) effect in galaxy clusters with a 4.9σ significance using the latest 217 GHz Planck map from data release 4. For the detection, we stacked the Planck map at the positions of 30 431 galaxy clusters from the Wen-Han-Liu (WHL) catalog. To align the sign of the kSZ signals, the line-of-sight velocities of galaxy clusters were estimated with a machine-learning approach, in which the relation between the galaxy distribution around a cluster and its line-of-sight velocity was trained through a convolutional neural network. To train our network, we used the simulated galaxies and galaxy clusters in the Magneticum cosmological hydrodynamic simulations. The trained model was applied to the large-scale distribution of the Sloan Digital Sky Survey galaxies to derive the line-of-sight velocities of the WHL galaxy clusters. Assuming a standard β-model for the intracluster medium, we obtained the gas mass fraction in R₅₀₀ to be f_gas, 500 = 0.09 ± 0.02 within the galaxy clusters with the average mass of M₅₀₀ ∼ 1.0 × 10¹⁴ h⁻¹ M_⊙.