Britton Lecture Series: Model theory and modeling

The third lecture by Thomas Scanlon entitled “Model Theory And Modeling: A Logician’s Approach To Parameter Identification”

Abstract:
Mathematical models are often described by systems of ODEs win which certain constants appear as parameters. Before using such a model, one ought to determine these parameters and the parameter identification problem asks for methods to do so from the outputs of experiments. Using the theory of canonical bases from model theoretic stability theory, we show that contrary to what had been expected, it is not always possible to recover the parameters from a single experiment, but that a bounded number of sufficiently independent experiments suffices. (The results are included in joint work with Alexey Ovchinnikov, Anand Pillay, and Gleb Pogudin.)

More info: https://www.math.mcmaster.ca/news-events/seminars-colloquia/britton-lectures.html