Memorial Union, 308

The Institute for AI & Computational Research is continuing the series of talks on AI/ML techniques and applications across various scientific domains. We envision the talks to strike a good balance between depth and breath. The goal of these talks will be to (i) introduce the particular AI/ML technique to fellow faculty and graduate students who have a basic understanding of deep learning and (ii) present a variety of applications in different domains without assuming deep domain knowledge. Each event will start with a 45 minute talk with 15 minutes for questions and a subsequent 30 minute networking session for brainstorming and further discussion. The speakers will be faculty from a number of URI colleges or nearby universities.



We are kicking off our fall seminar series with the talk below and are looking forward to see many of you at the seminar!


Speaker: Scott Field (Umass Dartmouth)
Date/Time/Location: Sept 18, 3pm, Memorial Union room 308.
Title: Physics-Informed Machine Learning for Modeling Orbital Dynamics in Binary Black Hole Systems

Abstract: Einstein’s field equations of general relativity are among the most challenging differential equations in physics. They predict that two black holes can collide, producing powerful distortions of spacetime—gravitational waves. The first direct detection of these waves in 2016 marked a historic milestone, recognized by a Nobel Prize.
In this talk, I will introduce the problem of binary black hole collisions and the role of gravitational waves. These waves can be studied either by numerically solving Einstein’s equations (the forward problem) or by analyzing experimental detections to infer their sources (the inverse problem). I will present a machine learning framework that discovers the governing dynamical model of binary black hole motion. Our approach begins with universal ordinary differential equations parameterized by feed-forward neural networks, and employs physics-informed constrained optimization to explore a space of plausible models. Applied to a variety of systems, this method enables forward extrapolation and automatically uncovers relativistic effects such as perihelion precession and radiation reaction. I will close by highlighting challenges and opportunities for extending these ideas to other areas of science and engineering.

Lectures / Presentations