Searching optimal scales for reconstructing cosmological initial conditions using convolutional neural networks

Abstract

Reconstructing the initial density field of the Universe from the late-time matter distribution is a nontrivial task with implications for understanding structure formation in cosmology, offering insights into early Universe conditions. Convolutional neural networks (CNNs) have shown promise in tackling this problem by learning the complex mapping from nonlinear evolved fields back to initial conditions. Here we investigate the effect of varying input sub-box size in single-input CNNs. We find that intermediate scales ($L_{\mathrm{sub}}\sim 152h^{-1}\mathrm{Mpc}$) strike the best balance between capturing local detail and global context, yielding the lowest validation loss and most accurate recovery across multiple statistical metrics. We then propose a dual-input model that combines two sub-boxes of different sizes from the same simulation volume. This model significantly improves reconstruction performance, especially on small scales over the best single-input case, despite utilizing the same parent simulation box. This demonstrates the advantage of explicitly incorporating multi-scale context into the network. Our results highlight the importance of input scale and network design in reconstruction tasks. The dual-input approach represents a simple yet powerful enhancement that leverages fixed input information more efficiently, paving the way for more accurate cosmological inference from large-scale structure surveys.

13 pages, 14 figures. Published version; revised following peer review

References in corpus (31)

• Seven-Year Wilkinson Microwave Anisotropy Probe (WMAP) Observations: Cosmological Interpretation
• The cosmological simulation code GADGET-2
• Observational Probes of Cosmic Acceleration
• Non-Gaussianity from Inflation: Theory and Observations
• Improving Cosmological Distance Measurements by Reconstruction of the Baryon Acoustic Peak
• Measuring D_A and H at z=0.35 from the SDSS DR7 LRGs using baryon acoustic oscillations
• Non-linear structure formation and the acoustic scale
• Reconstructing Baryon Oscillations: A Lagrangian Theory Perspective
• Getting Around Cosmic Variance
• Reconstructing baryon oscillations
• Towards optimal extraction of cosmological information from nonlinear data
• Iterative initial condition reconstruction
• Eulerian BAO Reconstructions and N-Point Statistics
• Nonlinear reconstruction
• Measuring the 2D Baryon Acoustic Oscillation signal of galaxies in WiggleZ: Cosmological constraints
• Analysis of the Kamionkowski-Loeb method of reducing cosmic variance with CMB polarization
• New method for initial density reconstruction
• Isobaric Reconstruction of the Baryonic Acoustic Oscillation
• Halo nonlinear reconstruction
• Baryon acoustic oscillations reconstruction using convolutional neural networks
• Reconstructing Cosmological Initial Conditions from Late-Time Structure with Convolutional Neural Networks
• Nonlinear reconstruction of redshift space distortions
• Effective cosmic density field reconstruction with convolutional neural network
• Increasing Fisher Information by Moving-Mesh Reconstruction
• Polarization Signal of Distant Clusters and Reconstruction of Primordial Potential Fluctuations
• Iterative removal of redshift space distortions from galaxy clustering
• Indra: a Public Computationally Accessible Suite of Cosmological $N$-body Simulations
• Understanding the Reconstruction of the Biased Tracer
• Primordial density and BAO reconstruction
• Measuring the cosmological density field twice: A novel test of dark energy using the CMB quadrupole
• Initial Conditions from Galaxies: Machine-Learning Subgrid Correction to Standard Reconstruction