Differentiable Physics: A Position Piece

2021/09/14 by Bharath Ramsundar, Ramsundar, Bharath, Dilip Krishnamurthy +3 · 11 citations
Biochemistry, Genetics and Molecular Biology · Computer Science · Mathematics · Physics and Astronomy · #Advanced Thermodynamics and Statistical Mechanics #Chemical Physics (physics.chem-ph) #Differentiable function #Economics #FOS: Computer and information sciences #FOS: Physical sciences #Machine Learning (cs.LG) #Mathematics #Model Reduction and Neural Networks #Physics #Position (finance) #Protein Structure and Dynamics #Pure mathematics #Theoretical physics #cs.LG #physics.chem-ph

paper · pdf · doi:10.48550/arxiv.2109.07573

published in arXiv (Cornell University) (Cornell University) · 12 pages, 1 figure

arxiv created 2021/09/14 · openalex publication_date 2021/09/14 · arxiv updated 2021/09/17 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28

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.

Citations

• Generalized Gradient Approximation Made Simple
• Fast Parallel Algorithms for Short-Range Molecular Dynamics
• Julia: A Fresh Approach to Numerical Computing
• TherML: Thermodynamics of Machine Learning
• Physics-informed neural networks: A deep learning framework for solving forward and inverse problems involving nonlinear partial differential equations
• Neural Stochastic Differential Equations: Deep Latent Gaussian Models in the Diffusion Limit
• PyTorch: An Imperative Style, High-Performance Deep Learning Library
• The Bottom-up Evolution of Representations in the Transformer: A Study with Machine Translation and Language Modeling Objectives
• DiffTaichi: Differentiable Programming for Physical Simulation
• The large learning rate phase of deep learning: the catapult mechanism
• Language Models are Few-Shot Learners
• Fourier Neural Operator for Parametric Partial Differential Equations
• ChemBERTa: Large-Scale Self-Supervised Pretraining for Molecular Property Prediction
• Extended Physics-Informed Neural Networks (XPINNs): A Generalized Space-Time Domain Decomposition Based Deep Learning Framework for Nonlinear Partial Differential Equations
• Deep learning based numerical approximation algorithms for stochastic partial differential equations
• Zero-Shot Text-to-Image Generation
• Learning Transferable Visual Models From Natural Language Supervision
• The Principles of Deep Learning Theory
• Highly accurate protein structure prediction with AlphaFold
• On the Opportunities and Risks of Foundation Models

Cited by

• Wall Shear Stress Reconstruction from Concentration: Differentiable Physics and Physics-Informed Neural Networks

Related

• A Review of Differentiable Simulators
• Differentiable programming across the PDE and Machine Learning barrier
• Differentiable Physics Simulations with Contacts: Do They Have Correct Gradients w.r.t. Position, Velocity and Control?
• Identifying Mechanical Models through Differentiable Simulations
• A Survey on Physics-based Differentiable Rendering