Mathematics Colloquia and Seminars

Return to Colloquia & Seminar listing

When parameterizing dynamical systems models of biological processes, we often use summary statistics (e.g., the mean) reported in experimental or observational studies. However, these summary statistics are abstractions, concealing variation occurring over space, time, or among individuals. Further, we know that the behavior of a nonlinear model using mean parameter values will differ from the mean model behavior if the parameter is instead treated as a random variable. Algae growing within polar sea ice provides an example of a system where extreme local heterogeneity in environmental conditions results in local heterogeneity in algal growth rates. Ignoring this and using a fixed, mean growth parameter to approximate regional dynamics can result in incorrect predictions of bloom timing and magnitude. Instead, algal growth rates at a given location should be treated as a random variable capturing the known heterogeneity. In this talk, I will provide an introduction to generalized polynomial chaos expansions, which provide elegant, computationally efficient methods for incorporating heterogeneous growth rates into standard algal bloom models, resulting in improved predictions of bloom dynamics. This method is broadly applicable for any system where local heterogeneity needs to be accounted for when considering aggregate dynamics over larger scales.