Learning to Predict the Cosmological Structure Formation

Abstract

Matter evolved under influence of gravity from minuscule density fluctuations. Non-perturbative structure formed hierarchically over all scales, and developed non-Gaussian features in the Universe, known as the Cosmic Web. To fully understand the structure formation of the Universe is one of the holy grails of modern astrophysics. Astrophysicists survey large volumes of the Universe and employ a large ensemble of computer simulations to compare with the observed data in order to extract the full information of our own Universe. However, to evolve trillions of galaxies over billions of years even with the simplest physics is a daunting task. We build a deep neural network, the Deep Density Displacement Model (hereafter D${}^3$M), to predict the non-linear structure formation of the Universe from simple linear perturbation theory. Our extensive analysis, demonstrates that D${}^3$M outperforms the second order perturbation theory (hereafter 2LPT), the commonly used fast approximate simulation method, in point-wise comparison, 2-point correlation, and 3-point correlation. We also show that D${}^3$M is able to accurately extrapolate far beyond its training data, and predict structure formation for significantly different cosmological parameters. Our study proves, for the first time, that deep learning is a practical and accurate alternative to approximate simulations of the gravitational structure formation of the Universe.

arXiv comment8 pages, 5 figures, 1 table

References in corpus (46)

• Adam: A Method for Stochastic Optimization
• Neural Message Passing for Quantum Chemistry
• WaveNet: A Generative Model for Raw Audio
• The 2dF Galaxy Redshift Survey: Spectra and redshifts
• Deep Speech 2: End-to-End Speech Recognition in English and Mandarin
• Solving the Quantum Many-Body Problem with Artificial Neural Networks
• SDSS-III: Massive Spectroscopic Surveys of the Distant Universe, the Milky Way Galaxy, and Extra-Solar Planetary Systems
• The Baryon Oscillation Spectroscopic Survey of SDSS-III
• Learning to Plan Chemical Syntheses
• GADGET: A code for collisionless and gasdynamical cosmological simulations
• Progressive Growing of GANs for Improved Quality, Stability, and Variation
• Improved Training of Wasserstein GANs
• End-to-End Training of Deep Visuomotor Policies
• The DESI Experiment Part I: Science,Targeting, and Survey Design
• How Filaments are Woven into the Cosmic Web
• The 6dF Galaxy Survey: Final Redshift Release (DR3) and Southern Large-Scale Structures
• Wide-Field InfrarRed Survey Telescope-Astrophysics Focused Telescope Assets WFIRST-AFTA 2015 Report
• Cosmology and Fundamental Physics with the Euclid Satellite
• The SDSS-IV extended Baryon Oscillation Spectroscopic Survey: Overview and Early Data
• Understanding Deep Convolutional Networks
• Galaxy And Mass Assembly (GAMA): end of survey report and data release 2
• Interaction Networks for Learning about Objects, Relations and Physics
• Bayesian physical reconstruction of initial conditions from large scale structure surveys
• Stochastic Video Generation with a Learned Prior
• Fast Automated Analysis of Strong Gravitational Lenses with Convolutional Neural Networks
• nbodykit: an open-source, massively parallel toolkit for large-scale structure
• Star-galaxy Classification Using Deep Convolutional Neural Networks
• FastPM: a new scheme for fast simulations of dark matter and halos
• Photometric Supernova Classification With Machine Learning
• CMU DeepLens: Deep Learning For Automatic Image-based Galaxy-Galaxy Strong Lens Finding
• DeepSphere: Efficient spherical Convolutional Neural Network with HEALPix sampling for cosmological applications
• Toward Controlled Generation of Text
• Lagrangian theory of gravitational instability of Friedman-Lemaitre cosmologies - generic third-order model for non-linear clustering
• Learning to Poke by Poking: Experiential Learning of Intuitive Physics
• The Zeldovich approximation
• The Initial Conditions of the Universe from Constrained Simulations
• Computing the Three-Point Correlation Function of Galaxies in $\mathcal{O}(N^2)$ Time
• Learning Visual Predictive Models of Physics for Playing Billiards
• TreePM: A code for Cosmological N-Body Simulations
• DeepCMB: Lensing Reconstruction of the Cosmic Microwave Background with Deep Neural Networks
• A volumetric deep Convolutional Neural Network for simulation of mock dark matter halo catalogues
• A New Parallel P3M Code for Very Large-Scale Cosmological Simulations
• Exploring the posterior surface of the large scale structure reconstruction
• Helmholtz Decomposition of the Lagrangian Displacement
• CosmoFlow: Using Deep Learning to Learn the Universe at Scale
• Biased Tracers in Redshift Space in the EFT of Large-Scale Structure

Cited by in corpus (120)

• Machine learning and the physical sciences
• Learning to Simulate Complex Physics with Graph Networks
• The Quijote simulations
• The CAMELS project: Cosmology and Astrophysics with MachinE Learning Simulations
• The CosmoVerse White Paper: Addressing observational tensions in cosmology with systematics and fundamental physics
• Large-scale dark matter simulations
• Euclid preparation: IX. EuclidEmulator2 -- Power spectrum emulation with massive neutrinos and self-consistent dark energy perturbations
• Exploring the Early Universe with Gaia and THEIA
• The Websky Extragalactic CMB Simulations
• The DAWES review 10: The impact of deep learning for the analysis of galaxy surveys
• AI-assisted super-resolution cosmological simulations
• What can Machine Learning tell us about the background expansion of the Universe?
• Painting with baryons: augmenting N-body simulations with gas using deep generative models
• Solving inverse-PDE problems with physics-aware neural networks
• The GIGANTES dataset: precision cosmology from voids in the machine learning era
• The CAMELS Multifield Dataset: Learning the Universe's Fundamental Parameters with Artificial Intelligence
• Super-resolution emulator of cosmological simulations using deep physical models
• The CAMELS project: public data release
• deep21: a Deep Learning Method for 21cm Foreground Removal
• Fast and Credible Likelihood-Free Cosmology with Truncated Marginal Neural Ratio Estimation
• Parameters Estimation for the Cosmic Microwave Background with Bayesian Neural Networks
• CARPool: fast, accurate computation of large-scale structure statistics by pairing costly and cheap cosmological simulations
• FlowPM: Distributed TensorFlow Implementation of the FastPM Cosmological N-body Solver
• Field Level Neural Network Emulator for Cosmological N-body Simulations
• A Hybrid Deep Learning Approach to Cosmological Constraints From Galaxy Redshift Surveys
• Lossless, Scalable Implicit Likelihood Inference for Cosmological Fields
• Gravitational wave population inference with deep flow-based generative network
• Learning effective physical laws for generating cosmological hydrodynamics with Lagrangian Deep Learning
• From Dark Matter to Galaxies with Convolutional Networks
• Painting halos from cosmic density fields of dark matter with physically motivated neural networks
• Removing Astrophysics in 21 cm maps with Neural Networks
• AI-assisted super-resolution cosmological simulations II: Halo substructures, velocities and higher order statistics
• The Role of Machine Learning in the Next Decade of Cosmology
• HIGAN: Cosmic Neutral Hydrogen with Generative Adversarial Networks
• Predicting dark matter halo formation in N-body simulations with deep regression networks
• Machine Learning for Observational Cosmology
• Neural network reconstructions for the Hubble parameter, growth rate and distance modulus
• Baryon acoustic oscillations reconstruction using convolutional neural networks
• Simulation-based inference of dynamical galaxy cluster masses with 3D convolutional neural networks
• NECOLA: Towards a Universal Field-level Cosmological Emulator
• Introduction to Latent Variable Energy-Based Models: A Path Towards Autonomous Machine Intelligence
• Nonlinear 3D Cosmic Web Simulation with Heavy-Tailed Generative Adversarial Networks
• Observational cosmology with Artificial Neural Networks
• Augmenting astrophysical scaling relations with machine learning: application to reducing the Sunyaev-Zeldovich flux-mass scatter
• HInet: Generating neutral hydrogen from dark matter with neural networks
• Reconstructing the cosmological density and velocity fields from redshifted galaxy distributions using V-net
• Bayesian Inference of Initial Conditions from Non-Linear Cosmic Structures using Field-Level Emulators
• A thorough investigation of the prospects of eLISA in addressing the Hubble tension: Fisher Forecast, MCMC and Machine Learning
• MillimeterDL: Deep Learning Simulations of the Microwave Sky
• Cosmic Velocity Field Reconstruction Using AI
• The bias of dark matter tracers: assessing the accuracy of mapping techniques
• Measuring the Hubble Constant with cosmic chronometers: a machine learning approach
• (DarkAI) Mapping the large-scale density field of dark matter using artificial intelligence
• Identification of point sources in gamma rays using U-shaped convolutional neural networks and a data challenge
• Investigating Cosmological GAN Emulators Using Latent Space Interpolation
• One simulation to have them all: performance of the Bias Assignment Method against N-body simulations
• Fast and Accurate Non-Linear Predictions of Universes with Deep Learning
• The cosmic web connection to the dark matter halo distribution through gravity
• Fast and realistic large-scale structure from machine-learning-augmented random field simulations
• Deep learning for Sunyaev-Zel'dovich detection in Planck
• Non-separable Covariance Kernels for Spatiotemporal Gaussian Processes based on a Hybrid Spectral Method and the Harmonic Oscillator
• Gradient boosting decision trees classification of blazars of uncertain type in the fourth Fermi-LAT catalog
• A New Emulated Monte Carlo Radiative Transfer Disk-Wind Model: X-Ray Accretion Disk-wind Emulator -- XRADE
• Teaching neural networks to generate Fast Sunyaev Zel'dovich Maps
• Large-scale structures in the $\Lambda$CDM Universe: network analysis and machine learning
• Improving constraints on primordial non-Gaussianity using neural network based reconstruction
• Optimizing simulation parameters for weak lensing analyses involving non-Gaussian observables
• Modeling assembly bias with machine learning and symbolic regression
• Eliminating polarization leakage effect for neutral hydrogen intensity mapping with deep learning
• Machine Learning and Cosmology
• Halo assembly bias from a deep learning model of halo formation
• AI-assisted super-resolution cosmological simulations III: Time evolution
• How do galaxy properties affect void statistics?
• Deep Forest: Neural Network reconstruction of the Lyman-alpha forest
• Augmenting Differentiable Simulators with Neural Networks to Close the Sim2Real Gap
• Efficient exploration and calibration of a semi-analytical model of galaxy formation with deep learning
• Bayesian Control Variates for optimal covariance estimation with pairs of simulations and surrogates
• Emulating Sunyaev-Zeldovich Images of Galaxy Clusters using Auto-Encoders
• Galaxy stellar and total mass estimation using machine learning
• PINION: Physics-informed neural network for accelerating radiative transfer simulations for cosmic reionization
• Deep learning insights into cosmological structure formation
• Learning neutrino effects in Cosmology with Convolutional Neural Networks
• Dark energy reconstruction analysis with artificial neural networks: Application on simulated Supernova Ia data from Rubin Observatory
• Cosmic Cartography: Bayesian reconstruction of the galaxy density informed by large-scale structure
• Simple lessons from complex learning: what a neural network model learns about cosmic structure formation
• Predicting 21cm-line map from Lyman $\alpha$ emitter distribution with Generative Adversarial Networks
• A field-level emulator for modified gravity
• A tomographic spherical mass map emulator of the KiDS-1000 survey using conditional generative adversarial networks
• BullFrog: Multi-step perturbation theory as a time integrator for cosmological simulations
• Perturbation Theory Remixed: Improved Nonlinearity Modeling beyond Standard Perturbation Theory
• Galaxy Spin Classification I: Z-wise vs S-wise Spirals With Chirality Equivariant Residual Network
• ScamPy -- A sub-halo clustering & abundance matching based Python interface for painting galaxies on the dark matter halo/sub-halo hierarchy
• A Machine Learning Approach to Correct for Mass Resolution Effects in Simulated Halo Clustering Statistics
• Psi-GAN: A power-spectrum-informed generative adversarial network for the emulation of large-scale structure maps across cosmologies and redshifts
• Peculiar Velocity Estimation from Kinetic SZ Effect using Deep Neural Networks
• COmoving Computer Acceleration (COCA): $N$-body simulations in an emulated frame of reference
• Point Source Detection with Fully-Convolutional Networks: Performance in Realistic Simulations
• Secondary halo bias through cosmic time II: Reconstructing halo properties using clustering information
• Learning the Universe: Learning to Optimize Cosmic Initial Conditions with Non-Differentiable Structure Formation Models
• Learning from learning machines: a new generation of AI technology to meet the needs of science
• kSZ Pairwise Velocity Reconstruction with Machine Learning
• Quasar Detection using Linear Support Vector Machine with Learning From Mistakes Methodology
• Improving the Accuracy of Halo Mass Based Statistics For Fast Approximate N-body Simulations
• Map Reconstruction of radio observations with Conditional Invertible Neural Networks
• Deep learning of material transport in complex neurite networks
• Deconfusing intensity maps with neural networks
• Success of Machine Learning algorithms in Dynamical Mass Measurements of Galaxy Clusters
• dm2gal: Mapping Dark Matter to Galaxies with Neural Networks
• Towards an optimal extraction of cosmological parameters from galaxy cluster surveys using convolutional neural networks
• Astro2020 Science White Paper: The Next Decade of Astroinformatics and Astrostatistics
• Cosmic velocity, density and halo mass function: Insights from deep learning
• Improving cosmological covariance matrices with machine learning
• Seeing Wiggles without Seeing Wiggles: BAO Recovery in 21 cm Intensity Mapping with Deep Learning
• Cosmological Analysis with Calibrated Neural Quantile Estimation and Approximate Simulators
• Cosmological constraints from the density gradient weighted correlation function
• Multi-frequency point source detection with fully convolutional networks: Performance in realistic microwave sky simulations
• nuGAN: Generative Adversarial Emulator for Cosmic Web with Neutrinos
• Predicting large scale cosmological structure evolution with generative adversarial network-based autoencoders
• DeepVoid: A Deep Learning Void Detector
• Multigrid Solver With Super-Resolved Interpolation