DATE:

Friday, April 28, 2023

TIME:

3:30 PM

LOCATION:

GMCS 314

SPEAKER:

Dr. Dmitrij Rappoport, Chemistry, University of CA, Irvine

ABSTRACT:

Complex chemical reactions are ubiquitous in chemistry but present significant challenges for computational exploration because of the high dimensionality of the corresponding reactive potential energy surface (PES) with many energy minima connected by transition states. In this talk, I will discuss two aspects, in which discrete representation of PESs as reaction networks makes their exploration easier. Reaction networks form discrete metric spaces of low dimensionality, about one order of magniture lower than that of the underlying PESs. Additionally, reaction networks define probability spaces, from which paths and subgraphs may be sampled. This opens new possibilities for constructing bias-free datasets of chemical reactions, for example, for assessing the accuracy of quantum chemical methods or for training machine learning models.

HOST:

Ricardo Carretero

VIDEO: