Molecules Represented as Graphs: A Key to Accelerating Molecular Simulations and Inverse Design
Feb 17, 2023
2:30PM to 3:30PM
Date(s) - 17/02/2023
2:30 pm - 3:30 pm
Prof. Rodrigo Vargas-Hernandez
Department of Chemistry & Chemical Biology, McMaster University
The simulation and inverse design of molecular systems are critical in many industrial processes, yet remain challenging tasks. Representing molecules as graphs and incorporating modern machine-learning tools offer a promising solution for acceleration. In this talk, we will introduce graph neural networks and differentiable molecular graphs as powerful tools for simulating molecular interactions and optimizing molecular structures for specific tasks.
Graph neural networks for simulating molecular systems provide new insights into the complex relationships between catalysts and reacting molecules.
We also introduce differentiable molecular graphs where we optimize the atomic compositions towards a target molecular property.
Join us to learn how the innovative use of graph representations is revolutionizing molecular simulations.
Rodrigo Vargas-Hernandez graduated from The National Autonomous University of Mexico in 2013. He did his Ph.D. under the supervision of prof. Krems in Vancouver BC where he discovers the field of machine learning. His first PDF was with prof. Brumer at Univ. of Toronto followed by a second in prof. Aspuru-Guzik, also at Univ. of Toronto. He was an affiliate PDF at the Vector Institute during his PDFs. In his free time, he likes to cycle and discover coffee places with tasty snacks.
In-Person: ABB 102