Building high accuracy emulators for scientific simulations with deep neural architecture search

Abstract

Computer simulations are invaluable tools for scientific discovery. However, accurate simulations are often slow to execute, which limits their applicability to extensive parameter exploration, large-scale data analysis, and uncertainty quantification. A promising route to accelerate simulations by building fast emulators with machine learning requires large training datasets, which can be prohibitively expensive to obtain with slow simulations. Here we present a method based on neural architecture search to build accurate emulators even with a limited number of training data. The method successfully accelerates simulations by up to 2 billion times in 10 scientific cases including astrophysics, climate science, biogeochemistry, high energy density physics, fusion energy, and seismology, using the same super-architecture, algorithm, and hyperparameters. Our approach also inherently provides emulator uncertainty estimation, adding further confidence in their use. We anticipate this work will accelerate research involving expensive simulations, allow more extensive parameters exploration, and enable new, previously unfeasible computational discovery.

References in corpus (7)

• Fast and Accurate Modeling of Molecular Atomization Energies with Machine Learning
• By-passing the Kohn-Sham equations with machine learning
• Laboratory evidence of dynamo amplification of magnetic fields in a turbulent plasma
• Galaxy clustering in the NEWFIRM Medium Band Survey: the relationship between stellar mass and dark matter halo mass at 1 < z < 2
• Cosmic Emulation: Fast Predictions for the Galaxy Power Spectrum
• Inverse Problem Instabilities in Large-Scale Plasma Modelling
• The galaxy-halo connection in the VIDEO Survey at 0.5<z<1.7

Cited by in corpus (10)

• Deep physical neural networks enabled by a backpropagation algorithm for arbitrary physical systems
• The data-driven future of high energy density physics
• Optimized and autonomous machine learning framework for characterizing pores, particles, grains and grain boundaries in microstructural images
• Accurate Machine Learning Atmospheric Retrieval via a Neural Network Surrogate Model for Radiative Transfer
• $\xi$-torch: differentiable scientific computing library
• Predicting Localized Primordial Star Formation with Deep Convolutional Neural Networks
• Opportunistic Emulation of Computationally Expensive Simulations via Deep Learning
• Reducing the Long Tail Losses in Scientific Emulations with Active Learning
• Designing Machine Learning Surrogates using Outputs of Molecular Dynamics Simulations as Soft Labels
• Robust posterior inference when statistically emulating forward simulations