STEM Speaker Series: Can a machine learn chemistry?
Tuesday, October 29 2024 at 1:00 PM EDT to
Tuesday, October 29 2024 at 2:00 PM EDT
Special Collections Seminar Room, E-2340, second floor of the Melville Library
Description
We live in a new scientific paradigm: the Big Data era, in which a lot of data is available for almost anything. In this new paradigm, the driving force is to use data directly to learn about chemical and physics systems employing artificial intelligence. This paradigm has proven helpful in simulating realistic physical, biological, and chemical models, yielding impressive results. Similarly, the insight gained in these situations can be used to improve our understanding of fundamental processes. In that regard, we want to answer the question: Can a machine learn chemistry? The answer to this question is still debatable, but we will show our ideas and methods to find the answer. Our group uses a bottom-up approach, starting with simple systems and then increasing the level of complexity. The simplest molecules, diatomics (two atoms bound to each other), will be the first system under consideration. Specifically, we will show that it is possible to predict molecular properties of diatomic molecules (spectroscopic constants and dipole moments) from atomic ones. Indeed, the level of accuracy is nearly as good as the gold standard in theoretical chemistry. Next, we will discuss our results on predicting atom-diatom reactions, showing that it is possible to predict the outcome of a reaction across the chemical space. Finally, we will discuss other avenues and work in progress in our group.