Skip to Content
PhysicsSimulation & Surrogates

Simulation & Surrogates

ML-accelerated physics simulations — surrogate models for PDEs, CFD, turbulence, and weather.

Tools Datasets Resources

Tools

Physics-Informed Neural Networks

PINNs Physics-informed neural networks DeepONet Deep operator networks Fourier Neural Operator Learning in Fourier space SINDy Sparse identification of dynamical systems NVIDIA PhysicsNeMo Framework for physics-ML models

Neural Differential Equations

Neural ODEs Continuous-depth neural networks Hamiltonian NNs Physics-preserving neural networks Universal Differential Equations Combining DEs with ML

SciML Software

torchdiffeq PyTorch neural ODEs DeepXDE Deep learning for scientific computing pysindy Sparse identification DifferentialEquations.jl Comprehensive DE solving (Julia) NeuralPDE.jl Physics-informed neural networks (Julia)

Books & Courses

Parallel Computing and SciML Chris Rackauckas Data-Driven Science and Engineering Brunton & Kutz

Datasets

PDEBench Comprehensive benchmark for PDE solving with ML PDEArena PDE modeling benchmark suite BLASTNet 744 full-domain samples of 3D turbulent flows JHTDB Johns Hopkins Turbulence Database Airfoil CFD 2D compressible flow simulations (6K samples) DrivAerNet 4,000 car meshes with aerodynamic data MeshGraphNets Data DeepMind simulation datasets PhiFlow Examples Physics simulation framework with data

Resources

SciML resources  Physics-informed neural networks  Berkeley Lab's curated scientific ML datasets 
Quantum SystemsSymbolic Regression