College of Science

Colloquium - Fall 2023

Monday, November 6, 2023
12:00pm - 1:15pm
Chapman Science Center, Rm S219
Lunch is provided!

Aerosol Modeling and Machine Learning in Chemistry

This talk presents an integrated approach that leverages supercomputers and machine learning to address challenges in computational chemistry. The first part employs high-performance computing to simulate the interfacial behavior of ions in atmospheric aerosols, a key factor in phenomena like ozone loss and acid rain. Molecular dynamics simulations were conducted on a ternary system involving surfactant dioctyl sodium sulfosuccinate (AOT), water, and isooctane solvent. The results reveal ionic layering in the interfacial region, providing critical insights for large- scale atmospheric models.

The second part introduces a supervised machine learning project aimed at identifying laboratory glassware. Implemented in an undergraduate course, the project utilized pre-trained neural networks— Inception-V1, Inception-V3, ResNet-50, and ResNet-101—for image classification. Training and testing were performed using the Wolfram Language, offering students hands-on experience in machine learning.

This presentation highlights the transformative role of high-performance computing and machine learning in both research and educational settings. It demonstrates how these computational tools can deepen our understanding of complex chemical systems and enhance the computational literacy of future scientists.