📝 SummarXiv

Abstract

A crucial feature of reaction-diffusion epidemic models is the incidence function, which characterizes disease transmission dynamics. Over the past few decades, many studies have investigated the behavior of such models under various incidence functions. However, the question of how to appropriately select a suitable incidence function remains largely unexplored. This paper addresses this issue by proposing an intuitive theoretical framework that recasts the original problem as determining the contributions of different incidence functions to the dynamics based on given observations. Specifically, the choice of an incidence function is linked to a weight assigned to it. Mathematically, this leads to a PDE-constrained optimization problem, where the objective is to identify the weights of a convex combination of multiple incidence functions that best approximate the experimental observations generated by the model. We first establish the Fr\'echet differentiability of the parameter-to-state operator and then derive the optimality conditions for the weights using a suitable adjoint problem. Finally, we illustrate our approach with a numerical example based on the Landweber iteration algorithm.