AI-assisted super-resolution cosmological simulations

Abstract

Cosmological simulations of galaxy formation are limited by finite computational resources. We draw from the ongoing rapid advances in Artificial Intelligence (specifically Deep Learning) to address this problem. Neural networks have been developed to learn from high-resolution (HR) image data, and then make accurate super-resolution (SR) versions of different low-resolution (LR) images. We apply such techniques to LR cosmological N-body simulations, generating SR versions. Specifically, we are able to enhance the simulation resolution by generating 512 times more particles and predicting their displacements from the initial positions. Therefore our results can be viewed as new simulation realizations themselves rather than projections, e.g., to their density fields. Furthermore, the generation process is stochastic, enabling us to sample the small-scale modes conditioning on the large-scale environment. Our model learns from only 16 pairs of small-volume LR-HR simulations, and is then able to generate SR simulations that successfully reproduce the HR matter power spectrum to percent level up to $16h^{-1}\mathrm{Mpc}$, and the HR halo mass function to within $10\%$ down to $10^{11}{M}_{\odot }$. We successfully deploy the model in a box 1000 times larger than the training simulation box, showing that high-resolution mock surveys can be generated rapidly. We conclude that AI assistance has the potential to revolutionize modeling of small-scale galaxy formation physics in large cosmological volumes.

10 pages, 8 figures, matches PNAS accepted version

References in corpus (5)

• Conditional Generative Adversarial Nets
• Learning effective physical laws for generating cosmological hydrodynamics with Lagrangian Deep Learning
• From Dark Matter to Galaxies with Convolutional Networks
• A unified framework for 21cm tomography sample generation and parameter inference with Progressively Growing GANs
• The Completed SDSS-IV Extended Baryon Oscillation Spectroscopic Survey: GLAM-QPM mock galaxy catalogs for the Emission Line Galaxy Sample

Cited by in corpus (46)

• Large-scale dark matter simulations
• The DAWES review 10: The impact of deep learning for the analysis of galaxy surveys
• The CAMELS Multifield Dataset: Learning the Universe's Fundamental Parameters with Artificial Intelligence
• The CAMELS project: public data release
• Field Level Neural Network Emulator for Cosmological N-body Simulations
• Learning effective physical laws for generating cosmological hydrodynamics with Lagrangian Deep Learning
• AI-assisted super-resolution cosmological simulations II: Halo substructures, velocities and higher order statistics
• Machine Learning for Observational Cosmology
• Accurate predictions from small boxes: variance suppression via the Zel'dovich approximation
• The evolution of barred galaxies in the EAGLE simulations
• NECOLA: Towards a Universal Field-level Cosmological Emulator
• Augmenting astrophysical scaling relations with machine learning: application to reducing the Sunyaev-Zeldovich flux-mass scatter
• From EMBER to FIRE: predicting high resolution baryon fields from dark matter simulations with Deep Learning
• A unified accretion disc model for supermassive black holes in galaxy formation simulations: method and implementation
• MillimeterDL: Deep Learning Simulations of the Microwave Sky
• A survey of machine learning-based physics event generation
• Fast and realistic large-scale structure from machine-learning-augmented random field simulations
• Modeling assembly bias with machine learning and symbolic regression
• Machine Learning and Cosmology
• Galaxies and Halos on Graph Neural Networks: Deep Generative Modeling Scalar and Vector Quantities for Intrinsic Alignment
• AI-assisted super-resolution cosmological simulations III: Time evolution
• Deep Forest: Neural Network reconstruction of the Lyman-alpha forest
• Deep learning insights into cosmological structure formation
• Classification algorithms applied to structure formation simulations
• Simple lessons from complex learning: what a neural network model learns about cosmic structure formation
• Foreground Removal of CO Intensity Mapping Using Deep Learning
• Black Hole Weather Forecasting with Deep Learning: A Pilot Study
• MF-Box: Multi-fidelity and multi-scale emulation for the matter power spectrum
• Evaluating the feasibility of interpretable machine learning for globular cluster detection
• A Machine Learning Approach to Correct for Mass Resolution Effects in Simulated Halo Clustering Statistics
• Astrophysics with the Laser Interferometer Space Antenna
• COmoving Computer Acceleration (COCA): $N$-body simulations in an emulated frame of reference
• Properties beyond mass for unresolved haloes across redshift and cosmology using correlations with local halo environment
• Stochastic Super-resolution of Cosmological Simulations with Denoising Diffusion Models
• Machine-guided Exploration and Calibration of Astrophysical Simulations
• Reconstructing the mid-infrared spectra of galaxies using ultraviolet to submillimeter photometry and Deep Generative Networks
• A Conditional Abundance Matching Method of Extending Simulated Halo Merger Trees to Resolve Low-Mass Progenitors and Sub-halos
• Convolutional Deep Denoising Autoencoders for Radio Astronomical Images
• Astrobiology in the Time of Artificial Intelligence
• Dreaming up scale invariance via inverse renormalization group
• Super-resolution simulation of the Fuzzy Dark Matter cosmological model
• Variational autoencoder for generating realistic $N$-body simulations for dark matter halos
• Improving cosmological covariance matrices with machine learning
• nuGAN: Generative Adversarial Emulator for Cosmic Web with Neutrinos
• EarthGAN: Can we visualize the Earth's mantle convection using a surrogate model?
• Correlated Read Noise Reduction in Infrared Arrays Using Deep Learning