DATE: 14 November 2022
LOCATION: Maryam Mirzakhani Seminar Room at BCAM and Online
TIME: 15:00

ABSTRACT
The SARS-CoV-2 pandemic has demonstrated the need for mathematical tools to make decisions and cope with the global health crisis. Modeling the spatial dynamics and influencing associated political measures have shown to be crucial in this context. There is a need for models that describe high-diffusive connections, such as airways or highways. However, up to now, nonlocal behavior in epidemics has only been modeled via networks. In the scope of this talk, we will present a novel approach that links mixed-dimensional diffusive models and uses a SIR-type (susceptible, infected, recovered) compartment model. We will investigate the mixed formulation concerning well-posedness and derive stable finite element discretizations. Subsequently, we will employ some of these diffusive models for modeling the environment in reinforcement learning-based control based on stochastic policy gradient methods for thermal and disease transmission dynamics problems. Our findings will be highlighted by showing theoretical and numerical results. This work is joint work with Aditya Vasudevan, David Portillo (UPM), Ignacio Romero, and Maciej Haranczyk.

Link to the session available here