Fall 2026
Fall 2026
September 25, 2026 Ian McGahan (UC Math)
Wildlife Movement through Heterogeneous Landscapes
Coordinated Animal movement is a longstanding problem in wildlife ecology. Understanding the effect of habitat heterogeneity in population distribution is of particular concern as this is critical to managing invasive species, understanding the impact of habitat loss on endangered species, and mitigating wildlife disease spread. While advances in tracking technology (e.g., GPS collars, camera traps) produced a wealth of spatio-temporal information about individual movement, linking this individual-scale data to population-scale consequences is, at best, computationally challenging. Partial differential equations (PDE) offer an avenue to link these disparate scales; the link between random walks and diffusion-type equations providing a prime example. However, the separation in spatial and temporal scale between the individual data (usually meters and hours) and population level questions like disease spread (tens to hundreds of kilometers and years) can pose significant analytic and computational barriers. In this talk expect to find an introduction to population-scale wildlife movement modeling, exploration of some of the analytic and computational challenges that arise, an application to white-tailed deer in Southwestern Wisconsin, and a discussion of the implications of this work on studying the spread of chronic wasting disease.
September 11, 2026 Zainab Almutawa (UC Math)
Modeling Beta Cell Network Dynamics: Genetic Algorithm Insights into Coupling Conditions and Critical Desynchronization
Coordinated insulin secretion depends on the electrical coupling of pancreatic beta cells via gap junctions. Variations in cell excitability and disruptions in coupling can compromise this synchrony, potentially contributing to diabetes. In this work, we use a computational beta cell network model to investigate how changes in gap junction connectivity and inherent cellular excitability impact synchronization. Using the Master Stability Function, we derive a necessary coupling condition for network-wide synchronization. A genetic algorithm efficiently navigates the multidimensional parameter space to identify candidate sets that sustain robust synchronization under normal conditions. These same parameter sets lead to complete desynchronization when a critical cell is silenced via chloride current activation. Furthermore, a neural network classifier differentiates between parameter sets leading to synchronized versus desynchronized states, underscoring a switch-like transition when a key cell is inactivated. Statistical analyses highlight the influence of individual parameters on network behavior. These findings provide new insights into beta cell coordination and the underlying mechanisms of insulin secretion dysregulation in diabetes.
TBD Tamara Lorenz (UC Psychology & Engineering)
Behavior Modelling of Humans and Robots
In CAP, the Center for Cognition, Action, and Perception at UC, we study human perception, action, and cognition from an ecological perspective. Our core assumption is that there are no mental representations of the world in our brains, and that the brain neither solely controls the body nor merely processes information. Instead, we view the brain–body–environment as a coupled system in which, over time, relevant information emerges—that is, becomes discoverable. Given our own action capabilities, we may perceive this information as action-relevant and experience the scene before us as offering action opportunities. When modeling the resulting human behavior, we therefore assume that all relevant information is available in the environment and can, in principle, be discovered. The models themselves serve as potential explanations of observable behavior, showing how behavior can be grounded in the environment given the laws of physics and our biology/physiology. Interestingly, individual behavior changes when we interact with others due to mutual adaptation of actions, and this same phenomenon can be observed in interactions with active machines, such as robots. To make robots safe and intuitive to interact with—and to better understand the subtle, nudging influence they can have on our behavior—I observe and dynamically model human action and interaction. In my talk, I will introduce our approach to dynamical systems modeling and present examples of my previous work, along with some early-stage ideas for future projects.