Institute

LEarning-Accelerated Domain Science (LEADS) Institute

Conceptual graphic for power grid simulations using neural networks

Composite image by Shannon Colson | Pacific Northwest National Laboratory

Lockup branding for the LEADS Institute

The LEarning-Accelerated Domain Science (LEADS) Scientific Discovery through Advanced Computing (SciDAC) institute will introduce a paradigm shift into research by integrating scientific machine learning (SciML) directly into domain-specific challenges. By combining artificial intelligence (AI) with scientific computing, SciML can significantly accelerate scientific discovery though complex mappings that uncover and quantify phenomena which have proven formidable for non-ML approaches.

LEADS will employ a protocol to gracefully scale the SciML approach from simple examples to the full complexity of a domain science problem. Our team is uniquely positioned to address phenomena where traditional numerical methods may be insufficient and expand the Department of Energy’s capability in using machine learning for domain science.

The LEADS SciDAC institute is supported by the Department of Energy (DOE), Office of Science, Advanced Scientific Computing Research program.