A Physics-ML Framework for Research

Superior-fidelity simulations in science and engineering are commonly made use of in industrial, seismic, weather/weather, and existence sciences purposes. However, standard simulations keep on being computationally high priced and impractical for serious time apps. They are discretization dependent, which means they do not effortlessly assimilate either calculated or synthetic data from various sources. Thanks to speedy developments in AI for science and engineering complications, machine learning has assumed an vital complementary role in addressing the crucial gaps in the traditional solutions.

NVIDIA Modulus is a physics-based mostly machine finding out system that has a number of point out-of-the-artwork network architectures and facts, as well as PDE pushed AI procedures to resolve true planet science and engineering difficulties. Many efficiency attributes for both one and multi-GPU/node programs, as well as connectivity with many NVIDIA toolkits and technologies are available in Modulus. Illustrations and documentation are offered to assure seamless learning for students though the researchers can customize the framework as a result of many APIs.

This webinar will introduce you to apps of machine studying, many domains of science and engineering, as perfectly as a deep dive into the code implementation, education, alternative, and visualization areas of physics-ML workflow.

By attending this webinar, you will understand about:

• The device learning applications in science & engineering with physics-ML framework, NVIDIA Modulus.

• How you can prolong/modify Modulus to put into action your very own get the job done.

• The architecture and performance of Modulus, and effectiveness enhancements for information & physics pushed programs.

• How the Modulus framework integrates with other Nvidia toolkits and systems: PySDF (for geometry), DALI™ (for data loading), Triton™ (for inference), Omniverse™  system (for visualization).

Sign up for us just after the presentation for a live Q&A session with Jianjun Xu, Ph.D., Sr. Alternatives Architect, Amazon Net Solutions.