NVIDIA PhysicsNeMo is an open-source deep-learning framework for building, training, fine-tuning and inferring Physics AI models using state-of-the-art SciML methods for AI4science and engineering.
PhysicsNeMo provides python modules to compose scalable and optimized training and inference pipelines to explore, develop, validate and deploy AI models that combine physics knowledge with data, enabling real-time predictions.
Whether you are exploring the use of Neural operators, GNNs, or transformers or are interested in Physics-informed Neural Networks or a hybrid approach in between, PhysicsNeMo provides you with an optimized stack that will enable you to train your models at scale.
PhysicsNeMo User Guide
- Training and Inference recipe
- Logging and Checkpointing recipe
- Profiling Applications in PhysicsNeMo
- Performance
- Domain Parallelism
- Adding Physics-based Information
PhysicsNeMo API
- PhysicsNeMo Models
- PhysicsNeMo Datapipes
- PhysicsNeMo Metrics
- PhysicsNeMo Deploy
- PhysicsNeMo Distributed
ShardTensor- PhysicsNeMo Utils
- PhysicsNeMo Launch Logging
- PhysicsNeMo Launch Utils
Introductory examples for learning key ideas
- Fourier Neural Operater for Darcy Flow
- Physics Guided models for Darcy flow
- Lid Driven Cavity Flow using Purely Physics Driven Neural Networks (PINNs)
- MeshGraphNet for transient vortex shedding
- Adaptive Fourier Neural Operator (AFNO) for weather forecasting
- MeshGraphNet with Lagrangian mesh
- Learning the flow field of Stokes flow
Examples: CFD
- MeshGraphNet for transient vortex shedding
- AeroGraphNet for external aerodynamic evaluation
- DoMINO: Decomposable Multi-scale Iterative Neural Operator for External Aerodynamics
- Factorized Implicit Global Convolution Network
- XAeroNet: Scalable Neural Models for External Aerodynamics
- RNN for transient 2D Navier Stokes flow
- RNN for transient 3D Gray Scott system
- MeshGraphNet with Lagrangian mesh
- Nested Fourier Neural Operater for Darcy Flow
- Physics Guided models for Darcy flow
- Learning the flow field of Stokes flow
- Shallow Water Equations - Distributed GraphCast
- Temporal attention model in Mesh-Reduced space for transient vortex shedding
- Diffusion-based-Fluid-Super-resolution
- Thermal and airflow surrogate model for Datacenter design
Examples: Weather and Climate
- ERA5 Data Downloader and Converter
- GraphCast for weather forecasting
- Adaptive Fourier Neural Operator (AFNO) for weather forecasting
- Deep Learning Weather Prediction (DLWP) model for weather forecasting
- Deep Learning Weather Prediction (DLWP-HEALPIX) model for weather forecasting
- Diagnostic models in PhysicsNeMo (precipitation)
- Unified Recipe for Training Global Weather Forecasting Models
- Generative Correction Diffusion Model (CorrDiff) for Km-scale Atmospheric Downscaling
- StormCast: Kilometer-Scale Convection Allowing Model Emulation using Generative Diffusion Modeling
Examples: Healthcare
- MeshGraphNet for Reduced-Order cardiovascular simulations
- Fourier Neural Operator for Brain anomaly detection
Examples: Additive Manufacturing
Examples: Molecular Dynamics