Computational and Applied Mathematics Seminar
Fall 2026
Mondays from 3:20pm to 4:10p.m (in-person (Room Carver 401) or ZOOM/WebEx/Teams talks)
The CAM Seminar is organized in the ISU Mathematics Department. It brings speakers from inside and outside of ISU, raising issues and exchanging ideas on topics of current interest in the area of computational and applied mathematics.
- October 19
Title: Identification and Optimization of Chloride Storage from Green Power
Micah Coats, Iowa State University
Abstract: Chlorine production accounts for about 2% of Germany’s total consumption. A chlorine storage process would shift this power consumption away from fossil fuels towards green energy sources, which tend to be more sporadic and often need some sort of energy storage process to be effectively used. To that end, a novel chlorine fluid capable of absorbing and releasing chlorine gas has been produced. Scaling from the lab to industry is complex. We created a numerical simulation to model the fluid behavior, with a focus on parameter identification within the model. Specifically, this work focuses on developing a Python based simulation of species transport, reaction, and heat transfer in a 2D axisymmetric liquid phase reactor.
- October 26
Title: Coefficient of restitution for an elastic rod in a viscous medium
Scott Hansen, Iowa State University
Abstract: A damped wave equation is used to model the motion of an elastic rod in a viscous medium that bounces off a fixed foundation with a given initial velocity under the influence of gravity. An explicit solution formula is derived and the time of contact with the foundation is determined in terms of the initial velocity, damping coefficient and gravitational constant. Explicit expressions are obtained for the motion of the center of mass for the duration of the impact. From this information, the internal vibrational energy present at the end of the contact time is obtained. A corresponding definition of an energy-based coefficient of restitution is proposed and analyzed.
- November 02
Title:
Abstract:
- November 09
Title:
Abstract:
- November 16
Title:
Amit Kumar Sikder, Iowa State University
Abstract:
- November 30
Title:
Farjana Siddiqua, Iowa State University
Abstract:
- August 31
Title: Organizational Meeting
Abstract: We will discuss the scope of CAM seminar for 2026-2027
- September 14
Title: A Mean-Field Theory of Transformers: Coupled Data–Parameter Dynamics and Global Convergence of Training
Hailiang Liu, Iowa State University
Abstract: Transformers have become a central architecture in modern machine learning, yet their mathematical analysis remains challenging because of the strong coupling between high-dimensional data representations and the trainable parameters.
In this talk I will present a mean-field framework for analyzing transformer training in the large-width regime. The resulting model describes the coupled evolution of data representations and parameter distributions. We establish well-posedness of this nonlinear mean-field dynamics and, under suitable assumptions, prove global convergence of training toward a global minimizer. The framework provides a mathematical foundation for understanding transformer training and its connection to mean-field theories of neural networks.
- September 28 (3:20 PM in Carver 232)
Title: Structure-Preserving Discretization of the Poisson–Nernst–Planck Equations via the Onsager Principle
Qianyi Li, Iowa State University
Abstract: This talk presents a variational approach for designing energy-stable and positivity-preserving numerical schemes for the Poisson–Nernst–Planck (PNP) equations. By leveraging the Onsager variational principle, the governing equations are reformulated as a constrained minimization problem. Time and spatial discretizations preserve this structure, leading to a well-posed minimization problem with linear constraints. The resulting scheme guarantees positivity, mass conservation, and unconditional energy stability, and numerical tests demonstrate its accuracy, stability, and robustness. The talk will also briefly introduce ongoing work on variational neural transport for aggregation–diffusion equations. Based on the Onsager variational principle, the dynamics are formulated in terms of a velocity field and its associated continuity equation, leading to a constrained optimization problem. Preliminary numerical results will be presented, together with ongoing theoretical and numerical work.