UTC RAVE Alert

UTC Mathematical Biology Webinars

UTC Mathematical Biology Webinar

Supported by NSF LEAPS-MPS 2532311, PI: Dr. Xiunan Wang

As part of Dr. Xiunan Wang's NSF LEAPS-MPS program, she is organizing a Mathematical Biology Webinar Series to support learning and engagement with mathematical modeling in biological systems. Hosted online, the series is open to undergraduate and graduate students, educators, and others interested in the intersection of mathematics and biology. Featuring speakers from diverse backgrounds and career stages, the series promotes professional development and interdisciplinary exploration in an inclusive and accessible setting.

Spring 2026

Dr. Sifan Wang

  • Biography: Dr. Sifan Wang is a Postdoctoral Fellow at Yale University’s Institute for Foundations of Data Science. He earned his Ph.D. in Applied Mathematics & Computational Science from the University of Pennsylvania (2023), advised by Dr. Paris Perdikaris. His research focuses on building reliable learning-based methods for physical systems governed by partial differential equations.
  • Thurs., Feb. 5 at 11:00 a.m.
  • Zoom Link: https://tennessee.zoom.us/j/89726697808
    • Password: 260205
  • Title: Toward a "GPT" moment for scientific computing
  • Abstract: Foundation models such as ChatGPT have reshaped AI by learning reusable representations that transfer across tasks. This talk asks whether a similar shift is possible in scientific computing: moving beyond solvers for a single partial differential equation (PDE) toward foundation models for families of PDE-governed systems. A central obstacle is that high-fidelity PDE data are expensive—often requiring hours to millions of CPU-hours per simulation—making purely data-driven scaling impractical. I present a physics-first roadmap that replaces data scale with physical structure, using governing equations as supervision. I will first focus on the single-PDE setting and show how physics-informed neural networks (PINNs) can be made reliable by diagnosing and addressing key training pathologies, leading to substantial accuracy improvements and successful simulations of challenging problems including 3D turbulence. I will then extend physics supervision from learning individual PDE solutions to learning solution operators for parametric PDE families. I will introduce the framework of physics-informed DeepONet and improve its scalability with continuous vision transformers. Finally, I will discuss how these advances motivate a longer-term direction toward unified models that can generalize across heterogeneous PDEs. Together, these results provide practical and theoretical steps toward PDE foundation models, with implications for accelerated simulation, design and control in computational science and engineering.

Dr. Briana Abrahms

Dr. Hao Wang

Dr. Suzanne Robertson

Dr. Sebastian Stockmaier

Dr. Veronica Ciocanel

Mathematics

Mathematics