PhysicsNeMo Examples Catalog#
Explore our comprehensive collection of examples organized by topic. Use the filters below to find examples that match your interests.
2D Darcy Flow using Fourier Neural Operators
How to set up a data-driven model for a 2D Darcy flow using the Fourier Neural Operator (FNO) architecture inside of PhysicsNeMo.
View Example2D Darcy Flow using Fourier Neural Operators and Physics Losses
This example demonstrates physics informing of a data-driven model using two approaches.
View ExampleLid Driven Cavity Flow using PINNs
Purely physics-driven model for solving a Lid Driven Cavity (LDC) flow using PINNs.
View ExampleMeshGraphNet for transient vortex shedding
This example is a re-implementation of the DeepMind's vortex shedding example deepmind/deepmind-research in PyTorch.
View ExampleAdaptive Fourier Neural Operator (AFNO) for weather forecasting
Code used for FourCastNet: A Global Data-driven High-resolution Weather Model using Adaptive Fourier Neural Operators.
View ExampleMeshGraphNet with Lagrangian mesh
Example of MeshGraphNet for particle-based simulation, based on the Learning to Simulate work.
View ExampleLearning the flow field of Stokes flow
MeshGraphNet model to learn the flow field of Stokes flow and further improve the accuary of the model predictions by physics-informed inference
View ExampleVirtual Foundary GraphNet
Graph-based deep learning approach to predict the part deformation
View ExampleAeroGraphNet for external aerodynamic evaluation
How to train the AeroGraphNet model for external aerodynamic analysis of both simplified (Ahmed body-type) and more realistic (DrivAerNet dataset) car geometries.
View ExampleDecomposable Multi-scale Iterative Neural Operator for External Aerodynamics
DoMINO is a local, multi-scale, point-cloud based model architecture to model large-scale physics problems such as external aerodynamics.
View ExampleFIGConvUNet for External Aerodynamics
FIGConvUNet is a novel architecture that can efficiently solve CFD problems for large 3D meshes and arbitrary input and output geometries.
View ExampleXAeroNet for External Aerodynamics
XAeroNet is a collection of scalable models for large-scale external aerodynamic evaluations. It consists of two models, XAeroNet-S and XAeroNet-V for surface and volume predictions, respectively.
View ExampleRNN for transient 2D Navier Stokes flow
This example uses recurrent neural networks for spatio-temporal prediction of the Navier Stokes flow.
View ExampleRNN for transient 3D Gray Scott system
This example uses recurrent neural networks for spatio-temporal prediction of the Gray-Scott reaction diffusion system.
View ExampleNested Fourier Neural Operater for Darcy Flow
This example demonstrates how to set up a data-driven model for a 2D Darcy flow using the Nested Fourier Neural Operator (FNO).
View ExampleShallow Water Equations - Distributed GraphCast
This example demonstrates how to leverage a distributed version of GraphCast to scale to larger Graph Neural Network (GNN) workloads.
View ExampleTemporal attention model in Mesh-Reduced space for transient vortex shedding
This example is an implementation of the paper “Predicting Physics in Mesh-reduced Space with Temporal Attention” in PyTorch.
View ExampleTransolver for Darcy Flow
This example demonstrates how to set up a data-driven model for a 2D Darcy flow using the Transolver inside of PhysicsNeMo.
View ExampleDiffusion-based-Fluid-Super-resolution
PyTorch implementation of "A Physics-informed Diffusion Model for High-fidelity Flow Field Reconstruction".
View ExampleThermal and airflow surrogate model for Datacenter design
Demonstrates the use of a Deep Learning model (3D UNet) for training a surrogate model for datacenter airflow to enable real-time datacenter design.
View ExampleMeshGraphNet for Reduced-Order cardiovascular simulations
Implements the one-dimensional Reduced-Order model based on MeshGraphNet presented in the paper Learning Reduced-Order Models for Cardiovascular Simulations with Graph Neural Networks.
View ExampleFourier Neural Operator for Brain anomaly detection
Fourier Neural Operator (FNO) for medical imaging.
View ExampleMolecular Dynamics using GNNs
MeshGraphNet model in PhysicsNeMo for developing a DL model for predicting forces/potential for a Lennard Jones System.
View ExampleERA5 Data Downloader and Converter
Tools for downloading ERA5 datasets via the Climate Data Store (CDS) API and processing them into formats suitable for machine learning
View ExampleGraphCast for weather forecasting
A re-implementation of the DeepMind’s GraphCast model in PhysicsNeMo.
View ExampleDLWP-HEALPIX model for weather forecasting
A re-implementation of the DLWP HEALPix model.
View ExampleDiagnostic models in PhysicsNeMo (precipitation)
This example contains code for training diagnostic models (models predicting an additional variable from the atmospheric state) using PhysicsNeMo.
View ExampleUnified Recipe for Training Global Weather Forecasting Models
This example demonstrates how to train a neural global weather forecast model.
View ExampleGenerative Correction Diffusion Model (CorrDiff) for Km-scale Atmospheric Downscaling
A cost-effective stochastic downscaling model, CorrDiff, is trained using high-resolution weather data and coarser ERA5 reanalysis.
View ExampleStormCast: Kilometer-Scale Convection Allowing Model Emulation using Generative Diffusion Modeling
This example demonstrates how to run training and simple inference for StormCast, a generative diffusion model designed to emulate NOAA’s High-Resolution Rapid Refresh (HRRR) model, a 3km operational CAM.
View Example1D Wave Equation (using only Physics)
This example, walks through the process of setting up a custom PDE in PhysicsNeMo Sym.
View Example2D Seismic Wave Propagation (using only Physics)
This example, extends the 1D wave equation example and solve a 2D seismic wave propagation problem commonly used in seismic surveying.
View ExampleCoupled Spring Mass ODE System (using only Physics)
This example, solve a system of coupled ordinary differential equations.
View ExampleTurbulent physics: Zero Equation Turbulence Model (using only Physics)
This example, walks through the process of adding a algebraic (zero equation) turbulence model to the PhysicsNeMo Sym simulations.
View ExampleScalar Transport: 2D Advection Diffusion (using only Physics)
This example, solves an advection-diffusion transport equation for temperature along with the Continuity and Navier-Stokes equation to model the heat transfer in a 2D flow.
View ExampleLinear Elasticity (using only Physics)
This exampleillustrates the linear elasticity implementation in PhysicsNeMo Sym.
View ExampleInverse Problem: Finding Unknown Coefficients of a PDE
Use PhysicsNeMo Sym to solve an inverse problem by assimilating data from observations.
View Example2D Darcy Flow using Fourier Neural Operators (via Symbolic Abstractions)
PhysicsNeMo Sym to set up a data-driven model for a 2D Darcy flow using the Fourier Neural Operator (FNO).
View Example2D Darcy Flow using Adaptive Fourier Neural Operators (via Symbolic Abstractions)
PhysicsNeMo Sym to set up a data-driven model for a 2D Darcy flow using the Adaptive Fourier Neural Operator (AFNO).
View Example2D Darcy Flow using Physics Informed Fourier Neural Operators (via Symbolic Abstractions)
PhysicsNeMo Sym to set up a data+physics model for a 2D Darcy flow using the Fourier Neural Operator (FNO).
View ExampleInterface Problem by Variational Method (using only Physics)
This example demonstrates the process of solving a PDE using the variational formulation.
View ExampleBlood Flow in Intracranial Aneurysm (using only Physics)
This example demonstrates how to train the PINNs to predict flow in a complex geometry imported as STL.
View ExampleMoving Time Window: Taylor Green Vortex Decay (using only Physics)
This example presents a moving time window approach to solve a complex transient Navier-Stokes problem.
View ExampleElectromagnetics: Frequency Domain Maxwell’s Equation (using only Physics)
This example demonstrates how to use PhysicsNeMo Sym to do the electromagnetic (EM) simulation.
View ExampleFully Developed Turbulent Channel Flow (using only Physics)
This example demonstrates the use of PINNs to solve a canonical turbulent flow in a 2D channel using two equation turbulence models and wall functions, without using any training data.
View ExampleTurbulence Super Resolution (via Symbolic Abstractions)
This example uses PhysicsNeMo Sym to train a super-resolution surrogate model for predicting high-fidelity homogeneous isotropic turbulence fields from filtered low-resolution observations provided by the Johns Hopkins Turbulence Database.
View ExampleConjugate Heat Transfer (using only Physics)
This example uses PhysicsNeMo Sym to study the conjugate heat transfer between the heat sink and the surrounding fluid.
View ExampleParameterized 3D Heat Sink (using only Physics)
This example walks through the process of simulating a parameterized problem using PhysicsNeMo Sym.
View ExampleHeat Transfer with High Thermal Conductivity (using only Physics)
This example discusses strategies that can be employed for handling conjugate heat transfer problems with higher thermal conductivities that represent more realistic materials.
View ExampleFPGA Heat Sink with Laminar Flow (using only Physics)
This example shows how some of the features in PhysicsNeMo Sym apply for a complicated FPGA heat sink design and solve the conjugate heat transfer.
View ExampleIndustrial Heat Sink CFD (using only Physics)
This example uses PhysicsNeMo Sym to conduct a thermal simulation of NVIDIA's NVSwitch heatsink.
View ExampleExplore Examples on GitHub#
Find more examples and contribute to our open-source projects on GitHub.