NVIDIA Modulus
NVIDIA Modulus
(Previously referred to as NVIDIA SimNet)
A Framework for Developing Physics Machine Learning Neural Network Models
If you’re an engineer, scientist, researcher, or student, using surrogate models is pervasive in your life’s work, from basic analysis workflows to building the most complex digital twin applications.
NVIDIA Modulus is a neural network framework that blends the power of physics and partial differential equations (PDEs) with AI to build more robust models for better analysis. Whether you’re looking to get started with AI-driven physics problems or designing digital twin models for complex non-linear, multi-physics systems, NVIDIA Modulus can support your work.
Benefits
Scalable Performance
Solves larger problems faster with accelerated linear algebra (XLA) and automatic mixed precision (AMP) support and multi-GPU/multi-node implementation.
Broad Applicability
Models multiple physics types in forward and inverse simulations with accuracy and convergence.
Fast Turnaround Time
Provides parameterized system representation that solves for multiple scenarios simultaneously, letting you train once to address multiple problems.
Easy to Adopt
Provides application programming interfaces (APIs) for implementing new physics and geometry and detailed user guide examples.
Modulus Multi-GPU/Multi-Node Performance
NVIDIA Modulus supports multi-GPU and multi-node scaling using Horovod. This allows for multiple processes, each targeting a single GPU, with collective communication using the NVIDIA Collective Communications Library (NCCL) and Message Passing Interface (MPI).
This plot shows the weak scaling performance of Modulus on a field-programmable gate array (FPGA) test problem running on up to 32 NVIDIA V100 Tensor Core GPUs in four NVIDIA DGX™-1 systems. The scaling efficiency from one to 32 GPUs is more than 85 percent.
Modulus Weak Scaling Across Multiple GPUs
Features
Modulus is a multi-physics framework that’s generalizable to multiple configurations enabled by parameterized geometry. With parameterized geometry, Modulus allows rapid design space exploration with single training for all configurations.
Novel Neural Network Architecture
Modulus provides a framework to model PDEs along with boundary and initial conditions. A key concept to solving the flow problems involves modeling the mass balance condition as a hard constraint as well as a global constraint. This improves accuracy and convergence characteristics. Additionally, for multi-physics problems spanning multiple domains, separate networks for the different physics with coupling at the domain interfaces work well.
Design Space Exploration
While traditional numerical solvers are designed to solve one configuration at a time, Modulus is able to work with multiple single geometries or parameterized geometry. The neural networks can be trained on multiple scenarios simultaneously and can evaluate each configuration in real time during inference. This allows the design space to be explored more efficiently.
Optimized for Multi-Physics Problems
Modulus is not only able to solve multiple-physics problems more efficiently with parameterized geometries but is also able to expand the scope of traditional simulations beyond currently solvable use cases. For example, the network can retain the knowledge gained during training and later solve the learned scenarios in real time. Similarly, the data assimilation and inverse problems that aren’t solved by the numerical solvers can be easily tackled by the neural networks.
For more details on Modulus (previously known as SimNet), please refer to NVIDIA SimNet: An AI-Accelerated Multi-Physics Simulation Framework. This paper reviews the neural network solver methodology, the Modulus architecture, and the features needed to effectively solve PDEs. It also includes real-world use cases that range from forward, parameterized, multi-physics simulations with turbulence and complex 3D geometries for industrial design optimization to inverse and data assimilation problems that aren’t addressed efficiently by traditional solvers.
What Others Are Saying
“[Modulus]’ clear APIs, clean and easily navigable code, environment, and hardware configurations well handled with dockers, scalability, ease of deployment, and the competent support team made it easy to adopt and has provided some very promising results. This has been great so far, and we look forward to using [Modulus] on problems with much larger dimensions.”
Cedric Frances, PhD Student, Stanford University
Using Physics-Informed Deep Learning for Transport in Porous Media
“[Modulus] is an AI-based physics simulation toolkit that has the potential to unlock amazing capabilities in industrial and scientific simulation.”
Christopher Lamb, VP of Computing Software, NVIDIA
“We believe that [Modulus] has some unique features like parameterized geometries for multi-physics problems and multi-GPU/multi-node neural network implementation. We are looking forward to incorporating [Modulus] in our research and teaching activities.”
Professor Hadi Meidani, Civil and Environmental Engineering, University of Illinois at Urbana-Champaign
Resources
- > Documentation
- > Forums
- > Webinar
- > Blogs
- > Using Hybrid Physics-Informed Neural Networks for Digital Twins in Prognosis and Health Management
- > Using Physics-Informed Deep Learning for Transport in Porous Media
- > NVIDIA Creates Framework for AI to Learn Physics
- > Accelerating Product Development with Physics-Informed Neural Networks and NVIDIA Modulus
- > Podcast - AI Physics Simulation Toolkit SimNet & PINNs
- > Demo - Siemens Energy HRSG Digital Twin Simulation Using NVIDIA Modulus and Omniverse