Deep learning for Sunyaev–Zel’dovich detection in Planck

Abstract

The Planck collaboration has extensively used the six Planck HFI frequency maps to detect the Sunyaev-Zel'dovich (SZ) effect with dedicated methods, e.g., by applying (i) component separation to construct a full sky map of the y parameter or (ii) matched multifilters to detect galaxy clusters via their hot gas. Although powerful, these methods may still introduce biases in the detection of the sources or in the reconstruction of the SZ signal due to prior knowledge (e.g., the use of the GNFW profile model as a proxy for the shape of galaxy clusters, which is accurate on average but not on individual clusters). In this study, we use deep learning algorithms, more specifically a U-Net architecture network, to detect the SZ signal from the Planck HFI frequency maps. The U-Net shows very good performance, recovering the Planck clusters in a test area. In the full sky, Planck clusters are also recovered, together with more than 18,000 other potential SZ sources, for which we have statistical hints of galaxy cluster signatures by stacking at their positions several full sky maps at different wavelengths (i.e., the CMB lensing map from Planck, maps of galaxy over-densities, and the ROSAT X-ray map). The diffuse SZ emission is also recovered around known large-scale structures such as Shapley, A399-A401, Coma, and Leo. Results shown in this proof-of-concept study are promising for potential future detection of galaxy clusters with low SZ pressure with this kind of approach, and more generally for potential identification and characterisation of large-scale structures of the Universe via their hot gas.

Constraining cosmology with a new all-sky Compton parameter map from the Planck PR4 data

“…The tSZ signal is subdominant relative to the CMB and other foreground emissions in the Planck frequency bands. Thus tailored component separation algorithms are required to reconstruct the tSZ map (i.e., Remazeilles et al 2011;Hurier et al 2013;Bourdin et al 2020;Bonjean 2020). We adopted the MILCA (Modified Internal Linear Combination Algorithm) (Hurier et al 2013) used for one of the Planck 𝑦-map reconstructions in Planck Collaboration (2016a) and mostly followed their reconstruction procedure (the difference in the procedure is summarized at the end of Sect.…”

“…The thermal Sunyaev-Zel'dovich (tSZ) effect (Sunyaev & Zeldovich 1972) is due to the inverse Compton scattering of cosmic microwave background (CMB) photons by hot electrons along the line of sight and in particular in clusters of galaxies. The tSZ effect has been measured by the Planck satellite, the Atacama Cosmology Telescope (ACT), and the South Pole Telescope (SPT) in a large field, and Compton parameter maps (i.e., Planck Collaboration 2014b, 2016aAghanim et al 2019;Madhavacheril et al 2020;Bleem et al 2021), referred to as 𝑦-map, have been constructed thanks to optimized component separation methods (i.e., Remazeilles et al 2011;Hurier et al 2013;Bourdin et al 2020;Bonjean 2020).…”

Search for Galaxy Cluster Candidates in the Cosmic Microwave Background Maps of the Planck Space Mission Using a Convolutional Neural Network Based on the Method of Tracing the Sunyaev–Zeldovich Effect

“…We aim to find such regions on the map. Unlike the methods developed earlier (Bonjean, 2020;Herranz et al, 2002;Melin 2006), in this paper we analyze the completeness of the obtained sample by comparing with simple Matched multi-filter (MMF1 (Herranz et al, 2002), MMF3 (Melin, 2006)) and PowellSnakes (PwS (Carvalho et al, 2012)) algorithms for finding objects. We also estimate the quality of the created model and the effect of ratio variations on the results of network learning, which may be useful when working with lower quality data.…”

“…We show that this approach works and can be used for data analysis in a search for objects with the SZeffect. The implemented approach supplements the existing algorithms (Bonjean, 2020) of searching for such objects, allowing to work with data in GLESP package format.…”

Encoding large-scale cosmological structure with generative adversarial networks

“…The recent advances in computer technology and in machine learning have prompted an increasing interest from the astronomical community in proposing machine learn-ing as an interesting alternative for the fast generation of mock simulations and mock data, or for image processing. The ever larger quantities and quality of astronomical data call for systematic approaches to properly interpret and extract the information that can be based on machinelearning techniques such as in Villaescusa-Navarro et al (2020), Schawinski et al (2018), or Bonjean (2020). Machine learning can also be used to produce density maps from large N-body simulations of dark matter (DM) (Rodríguez et al 2018;Feder et al 2020) in a computationally cheaper manner, to predict the effects of DM annihilation feedback on gas densities (List et al 2019), or to infer a mapping between the N-body and the hydrodynamical simulations without resorting to full simulations (Tröster et al 2019;Zamudio-Fernandez et al 2019).…”