Benchmarks and Explanations for Deep Learning Estimates of X-ray Galaxy Cluster Masses

Abstract

We evaluate the effectiveness of deep learning (DL) models for reconstructing the masses of galaxy clusters using X-ray photometry data from next-generation surveys. We establish these constraints using a catalogue of realistic mock eROSITA X-ray observations which use hydrodynamical simulations to model realistic cluster morphology, background emission, telescope response, and AGN sources. Using bolometric X-ray photon maps as input, DL models achieve a predictive mass scatter of ${\sigma }_{\ln M_{500c}}=17.8\%$, a factor of two improvements on scalar observables such as richness $N_{\mathrm{gal}}$, 1D velocity dispersion ${\sigma }_{\mathrm{v},1D}$, and photon count $N_{\mathrm{phot}}$ as well as a $32\%$ improvement upon idealised, volume-integrated measurements of the bolometric X-ray luminosity $L_X$. We then show that extending this model to handle multichannel X-ray photon maps, separated in low, medium, and high energy bands, further reduces the mass scatter to $16.2\%$. We also tested a multimodal DL model incorporating both dynamical and X-ray cluster probes and achieved marginal gains at a mass scatter of $15.9\%$. Finally, we conduct a quantitative interpretability study of our DL models and find that they greatly down-weight the importance of pixels in the centres of clusters and at the location of AGN sources, validating previous claims of DL modelling improvements and suggesting practical and theoretical benefits for using DL in X-ray mass inference.

14 pages, 9 figures, 3 tables, accepted in MNRAS

References in corpus (24)

• Five-Year Wilkinson Microwave Anisotropy Probe (WMAP) Observations: Cosmological Interpretation
• Testing X-ray Measurements of Galaxy Clusters with Cosmological Simulations
• eROSITA Science Book: Mapping the Structure of the Energetic Universe
• Virial Scaling of Massive Dark Matter Halos: Why Clusters Prefer a High Normalization Cosmology
• LoCuSS: Comparison of Observed X-ray and Lensing Galaxy Cluster Scaling Relations with Simulations
• XMM-Newton and Chandra Cross Calibration Using HIFLUGCS Galaxy Clusters: Systematic Temperature Differences and Cosmological Impact
• A Model for Multi-property Galaxy Cluster Statistics
• Active Galactic Nuclei and their Large-scale Structure: an eROSITA mock catalogue
• LoCuSS: Scaling relations between galaxy cluster mass, gas, and stellar content
• Morphology parameters: substructure identification in X-ray galaxy clusters
• Detection of anti-correlation of hot and cold baryons in galaxy clusters
• Galaxy cluster mass estimation with deep learning and hydrodynamical simulations
• Galaxy Cluster Mass Reconstruction Project: III. The impact of dynamical substructure on cluster mass estimates
• Nature versus Nurture: The curved spine of the galaxy cluster X-ray luminosity -- temperature relation
• A Deep Learning Approach to Infer Galaxy Cluster Masses from Planck Compton$-y$ parameter maps
• Using X-Ray Morphological Parameters to Strengthen Galaxy Cluster Mass Estimates via Machine Learning
• The Dynamical Mass of the Coma Cluster from Deep Learning
• SPIDERS: overview of the X-ray galaxy cluster follow-up and the final spectroscopic data release
• Aging Halos: Implications of the Magnitude Gap on Conditional Statistics of Stellar and Gas Properties of Massive Halos
• Sunyaev-Zel'dovich effect and X-ray scaling relations of galaxies, groups and clusters in the IllustrisTNG simulations
• The Importance of Being Interpretable: Toward An Understandable Machine Learning Encoder for Galaxy Cluster Cosmology
• A Machine Learning Approach to Enhancing eROSITA Observations
• The eROSITA Final Equatorial-Depth Survey (eFEDS): A Machine Learning Approach to Infer Galaxy Cluster Masses from eROSITA X-ray Images
• Unifying Sunyaev-Zel'dovich and X-ray predictions from clusters to galaxy groups: the impact of X-ray mass estimates on the $Y-M$ scaling relation

Cited by in corpus (8)

• The Three Hundred Project: Mapping The Matter Distribution in Galaxy Clusters Via Deep Learning from Multiview Simulated Observations
• Identifying Galaxy Cluster Mergers with Deep Neural Networks using Idealized Compton-y and X-ray maps
• kSZ Pairwise Velocity Reconstruction with Machine Learning
• A Generative Model for Realistic Galaxy Cluster X-ray Morphologies
• Deriving accurate galaxy cluster masses using X-ray thermodynamic profiles and graph neural networks
• Cluster optical depth and pairwise velocity estimation using machine learning
• Generating Galaxy Clusters Mass Density Maps from Mock Multiview Images via Deep Learning
• Estimating the triaxiality of massive clusters from 2D observables in MillenniumTNG with machine learning