A Physics-ML Framework for Analysis

Excessive-fidelity simulations in science and engineering are extensively utilized in industrial, seismic, climate/local weather, and life sciences purposes. Nonetheless, conventional simulations stay computationally costly and impractical for actual time purposes. They’re discretization dependent, that means they don’t simply assimilate both measured or artificial information from varied sources. On account of fast developments in AI for science and engineering issues, machine studying has assumed an essential complementary function in addressing the vital gaps within the conventional strategies.

NVIDIA Modulus is a physics-based machine studying platform that has a number of state-of-the-art community architectures and information, in addition to PDE pushed AI methods to unravel actual world science and engineering issues. Varied efficiency options for each single and multi-GPU/node programs, plus connectivity with a number of NVIDIA toolkits and applied sciences can be found in Modulus. Examples and documentation are offered to make sure seamless studying for college kids whereas the researchers can customise the framework by way of varied APIs.

This webinar will introduce you to purposes of machine studying, varied domains of science and engineering, in addition to a deep dive into the code implementation, coaching, answer, and visualization points of physics-ML workflow.

By attending this webinar, you’ll find out about:

• The machine studying purposes in science & engineering with physics-ML framework, NVIDIA Modulus.

• How one can lengthen/modify Modulus to implement your personal work.

• The structure and performance of Modulus, and efficiency enhancements for information & physics pushed programs.

• How the Modulus framework integrates with different Nvidia toolkits and applied sciences: PySDF (for geometry), DALI™ (for information loading), Triton™ (for inference), Omniverse™  platform (for visualization).

Be a part of us after the presentation for a reside Q&A session with Jianjun Xu, Ph.D., Sr. Options Architect, Amazon Internet Companies.