Project Grant 2227112

This National Science Foundation Project Grant of $418,678 awarded on August 15, 2022 will support the development of computational tools to model complex chemical reactions through dimensionality reduction and machine learning techniques. Under the federal Mathematical and Physical Sciences program (CFDA 47.049), Dmitrij Rappoport of the University of California, Irvine will create low-dimensional representations of potential energy surfaces using nonlinear dimensionality reduction and discretization methods. These computational models aim to separate reactive and nonreactive degrees of freedom to efficiently sample reaction mechanisms and predict chemical reactivities while avoiding limitations of full-dimensional modeling. As the prime awardee, UC Irvine will deliver new machine learning-based methods to complement existing transition state search tools and address high dimensionality challenges in computational discovery and optimization of chemical syntheses.