Differentiable Physics: A Position Piece

Abstract

Differentiable physics provides a new approach for modeling and understanding the physical systems by pairing the new technology of differentiable programming with classical numerical methods for physical simulation. We survey the rapidly growing literature of differentiable physics techniques and highlight methods for parameter estimation, learning representations, solving differential equations, and developing what we call scientific foundation models using data and inductive priors. We argue that differentiable physics offers a new paradigm for modeling physical phenomena by combining classical analytic solutions with numerical methodology using the bridge of differentiable programming.

arXiv comment12 pages, 1 figure

References in corpus (38)

• PyTorch: An Imperative Style, High-Performance Deep Learning Library
• Learning Transferable Visual Models From Natural Language Supervision
• Discovering governing equations from data: Sparse identification of nonlinear dynamical systems
• Language Models are Few-Shot Learners
• Crystal Graph Convolutional Neural Networks for an Accurate and Interpretable Prediction of Material Properties
• On the Opportunities and Risks of Foundation Models
• DeePMD-kit: A deep learning package for many-body potential energy representation and molecular dynamics
• Solving high-dimensional partial differential equations using deep learning
• Graph Networks as a Universal Machine Learning Framework for Molecules and Crystals
• Zero-Shot Text-to-Image Generation
• Fourier Neural Operator for Parametric Partial Differential Equations
• Group Equivariant Convolutional Networks
• ChemBERTa: Large-Scale Self-Supervised Pretraining for Molecular Property Prediction
• The Principles of Deep Learning Theory
• Flow-based generative models for Markov chain Monte Carlo in lattice field theory
• Equivariant flow-based sampling for lattice gauge theory
• Kohn-Sham equations as regularizer: building prior knowledge into machine-learned physics
• Evidence for supercritical behavior of high-pressure liquid hydrogen
• Machine learning on DNA-encoded libraries: A new paradigm for hit-finding
• State Predictive Information Bottleneck
• Sampling using $SU(N)$ gauge equivariant flows
• Machine learning based energy-free structure predictions of molecules (closed and open-shell), transition states, and solids
• Active learning of deep surrogates for PDEs: Application to metasurface design
• Supervised Learning with Quantum-Inspired Tensor Networks
• DeePKS: a comprehensive data-driven approach towards chemically accurate density functional theory
• Learning to learn with quantum neural networks via classical neural networks
• DiffTaichi: Differentiable Programming for Physical Simulation
• Turbulent scalar flux in inclined jets in crossflow: counter gradient transport and deep learning modelling
• The large learning rate phase of deep learning: the catapult mechanism
• A Simple Differentiable Programming Language
• Neural Stochastic Differential Equations: Deep Latent Gaussian Models in the Diffusion Limit
• Neural networks with redundant representation: detecting the undetectable
• Data-Driven Discovery of Coarse-Grained Equations
• TherML: Thermodynamics of Machine Learning
• The Bottom-up Evolution of Representations in the Transformer: A Study with Machine Translation and Language Modeling Objectives
• Deep learning based numerical approximation algorithms for stochastic partial differential equations
• Simulating Continuum Mechanics with Multi-Scale Graph Neural Networks
• Differentiable thermodynamic modeling

Cited by in corpus (1)

• Inverse molecular design and parameter optimization with Hückel theory using automatic differentiation