Uncertainty-Aware Neural Networks for Fuzzy Dark Matter Model Selection from \texorpdfstringx_{\rm HI}x_HI Measurements

2025/09/06 by Salmasi, Bahareh Soleimanpour, Hosseini, S. Mobina
#Cosmology and Nongalactic Astrophysics (astro-ph.CO) #FOS: Physical sciences

paper · doi:10.48550/arxiv.2509.05551

Abstract

The nature of dark matter remains a central question in cosmology, with fuzzy dark matter (FDM) models offering a compelling alternative to the cold dark matter (CDM) paradigm. We explore FDM scenarios by performing 21-cm simulations across a parameter space with \texorpdfstringf_{\rm FDM} ∈ [0.02, 0.10]f_FDM in [0.02, 0.10] and \texorpdfstringm_{\rm FDM} ∈ [10^-24, 10^-21] eVm_FDM in [10^-24, 10^-21] eV, obtaining global neutral hydrogen fractions (\texorpdfstringx_{\rm HI}x_HI) for each model. Observational \texorpdfstringx_{\rm HI}x_HI data and associated uncertainties from JWST are incorporated by estimating full probability density functions (PDFs) for both \texorpdfstringx_{\rm HI}x_HI and redshift \texorpdfstringzz using Bayesian inference with the No-U-Turn Sampler (NUTS), yielding non-Gaussian multivariate uncertainty distributions. A hybrid machine learning framework is then trained on these observational PDFs to learn both central values and correlated uncertainties in \texorpdfstringx_{\rm HI}x_HI and \texorpdfstringzz, iteratively refining its parameters in each training epoch through direct incorporation of the multivariate PDFs derived from observational constraints. We then compare the simulation outputs to the machine-learned observational trends to identify the most consistent models. Our results indicate that FDM models with \texorpdfstringm_{\rm FDM} ≃ 10^-22 eVm_FDM approx 10^-22 eV and \texorpdfstringf_{\rm FDM} ≃ 0.04f_FDM approx 0.04 best match current data, while lighter masses are strongly constrained. By integrating simulations and machine learning in an uncertainty-aware framework, this work explores the physics of the early Universe and guides future studies of 21-cm cosmology and reionization.

Citations

• Exploring the Quiescent Black Hole Population of Nearby Dwarf Galaxies with the Habitable World Observatory
• Boosting Supermassive Black Hole Growth in the Early Universe by Fuzzy Dark Matter Solitons
• Constraints on the fuzzy dark matter mass window from high-redshift observables
• High-redshift, small-scale tests of ultralight axion dark matter using Hubble and Webb galaxy UV luminosities
• Ultra-light dark matter explanation of NANOGrav observations
• Retrieving the 21-cm signal from the Epoch of Reionization with learnt Gaussian process kernels
• The NANOGrav 15 yr Data Set: Search for Signals from New Physics
• The NANOGrav 15 yr Data Set: Evidence for a Gravitational-wave Background
• A New Constraint on the Simulation of the Intergalactic Medium through the Evolution of the Neutral Hydrogen Fraction in the Epoch of Reionization
• JWST Measurements of Neutral Hydrogen Fractions and Ionized Bubble Sizes at z=7-12 Obtained with Lyα Damping Wing Absorptions in 27 Bright Continuum Galaxies
• JWST NIRSpec spectroscopy of the triply-lensed z = 10.17 galaxy MACS0647-JD
• THESAN-HR: Galaxies in the Epoch of Reionization in warm dark matter, fuzzy dark matter and interacting dark matter
• In-orbit Performance of the Near-infrared Spectrograph NIRSpec on the James Webb Space Telescope
• JWST Census for the Mass-Metallicity Star-Formation Relations at z=4-10 with the Self-Consistent Flux Calibration and the Proper Metallicity Calibrators
• Key Physical Processes in the Circumgalactic Medium
• Performance of NIRCam on JWST in Flight
• Spectroscopic confirmation of four metal-poor galaxies at z=10.3-13.2
• DeepLSS: breaking parameter degeneracies in large scale structure with\n deep learning analysis of combined probes
• Alleviating both H₀ and S₈ tensions: early dark energy lifts the CMB-lockdown on ultralight axion
• The coefficient of determination R-squared is more informative than SMAPE, MAE, MAPE, MSE and RMSE in regression analysis evaluation
• A review on the long short-term memory model
• A Survey of Convolutional Neural Networks: Analysis, Applications, and Prospects
• A Survey of Convolutional Neural Networks: Analysis, Applications, and Prospects
• 1D Convolutional Neural Networks and Applications: A Survey
• 1D convolutional neural networks and applications: A survey
• The Role of Machine Learning in the Next Decade of Cosmology
• Small-Scale Challenges to the <b><i>Λ</i></b>CDM Paradigm
• An overview of gradient descent optimization algorithms
• Recent Advances in Convolutional Neural Networks
• Recent advances in convolutional neural networks
• Empirical Evaluation of Gated Recurrent Neural Networks on Sequence Modeling
• Deep learning in neural networks: An overview
• GRB 130606A within a sub-DLA at redshift 5.91
• On the difficulty of training Recurrent Neural Networks
• The Cosmic Linear Anisotropy Solving System (CLASS) I: Overview
• Reionization and Cosmology with 21-cm Fluctuations
• Cosmology at low frequencies: The 21cm transition and the high-redshift Universe
• Implications for Cosmic Reionization from the Optical Afterglow Spectrum of the Gamma-Ray Burst 050904 at <i>z</i> = 6.3
• In the beginning: the first sources of light and the reionization of the universe
• The Structure and Dynamics of Cold Dark Matter Halos
• Long Short-Term Memory
• The Structure of Dark Matter Halos in Dwarf Galaxies
• Learning long-term dependencies with gradient descent is difficult

Related

• Cosmo-SPINN: Fuzzy Dark Matter Simulations with Physics-Informed Generative Networks
• Constraints on the fuzzy dark matter mass window from high-redshift observables
• No Catch-22 for Fuzzy Dark Matter: testing substructure counts and core sizes via high-resolution cosmological simulations
• Seeking New Physics in Cosmology with Bayesian Neural Networks: Dark Energy and Modified Gravity
• Fuzzy Gasoline: Cosmological hydrodynamical simulations of dwarf galaxy formation with Fuzzy Dark Matter