Stochastic Super-resolution of Cosmological Simulations with Denoising Diffusion Models

Abstract

In recent years, deep learning models have been successfully employed for augmenting low-resolution cosmological simulations with small-scale information, a task known as "super-resolution". So far, these cosmological super-resolution models have relied on generative adversarial networks (GANs), which can achieve highly realistic results, but suffer from various shortcomings (e.g. low sample diversity). We introduce denoising diffusion models as a powerful generative model for super-resolving cosmic large-scale structure predictions (as a first proof-of-concept in two dimensions). To obtain accurate results down to small scales, we develop a new "filter-boosted" training approach that redistributes the importance of different scales in the pixel-wise training objective. We demonstrate that our model not only produces convincing super-resolution images and power spectra consistent at the percent level, but is also able to reproduce the diversity of small-scale features consistent with a given low-resolution simulation. This enables uncertainty quantification for the generated small-scale features, which is critical for the usefulness of such super-resolution models as a viable surrogate model for cosmic structure formation.

arXiv comment9 pages, 8 figures, to be submitted to OJA, comments welcome

References in corpus (35)

• A Universal Density Profile from Hierarchical Clustering
• Denoising Diffusion Probabilistic Models
• Simulating the joint evolution of quasars, galaxies and their large-scale distribution
• Gaussian Error Linear Units (GELUs)
• Proceedings of the 29th International Conference on Machine Learning (ICML-12)
• Deep Unsupervised Learning using Nonequilibrium Thermodynamics
• Ten equivalent definitions of the fractional Laplace operator
• Cascaded Diffusion Models for High Fidelity Image Generation
• The Quijote simulations
• The Uchuu Simulations: Data Release 1 and Dark Matter Halo Concentrations
• AbacusSummit: A Massive Set of High-Accuracy, High-Resolution $N$-Body Simulations
• FastPM: a new scheme for fast simulations of dark matter and halos
• AI-assisted super-resolution cosmological simulations
• An adaptively refined phase-space element method for cosmological simulations and collisionless dynamics
• Painting with baryons: augmenting N-body simulations with gas using deep generative models
• Super-resolution emulator of cosmological simulations using deep physical models
• Image Super-Resolution via Iterative Refinement
• Painting halos from cosmic density fields of dark matter with physically motivated neural networks
• Probabilistic Mass Mapping with Neural Score Estimation
• AI-assisted super-resolution cosmological simulations II: Halo substructures, velocities and higher order statistics
• HIGAN: Cosmic Neutral Hydrogen with Generative Adversarial Networks
• Nonlinear 3D Cosmic Web Simulation with Heavy-Tailed Generative Adversarial Networks
• From EMBER to FIRE: predicting high resolution baryon fields from dark matter simulations with Deep Learning
• Quantifying resolution in cosmological N-body simulations using self-similarity
• Encoding large scale cosmological structure with Generative Adversarial Networks
• A black box for dark sector physics: Predicting dark matter annihilation feedback with conditional GANs
• A unified framework for 21cm tomography sample generation and parameter inference with Progressively Growing GANs
• Investigating Cosmological GAN Emulators Using Latent Space Interpolation
• Posterior samples of source galaxies in strong gravitational lenses with score-based priors
• Deep learning reconstruction of three-dimensional galaxy distributions with intensity mapping observations
• Predicting 21cm-line map from Lyman $\alpha$ emitter distribution with Generative Adversarial Networks
• Normalizing flows for random fields in cosmology
• A deep learning approach to halo merger tree construction
• Strong-Lensing Source Reconstruction with Denoising Diffusion Restoration Models
• Super-Resolution Emulation of Large Cosmological Fields with a 3D Conditional Diffusion Model

Cited by in corpus (1)

• Detecting Modeling Bias with Continuous Time Flow Models on Weak Lensing Maps