In many areas of science and industry, decisions are taken based on predictions of the evolution of a physical system’s state (concentration of polluants in the city, blood flux in an artery, population of neutrons in a nuclear core…). Some of these decisions may also need to be taken in real time based on continuous updates. Models & Measures is a research group financed by the Emergences Program of the Paris City Council which aims at developing approximation methods for estimating the state of a system.
So far, such problems have been tackled by interpolation techniques relying on data from measuring devices or by physical modeling (e.g., solution of partial differential equations). Both approaches lead to different yet incomplete descriptions of the system due to its is inherent complexity. This project explores the opportunity to couple both approaches in order to complement their strengths, especially by trying to reduce the intrisic high dimensionality of the problems.
Keywords: reduced modeling, high dimensional problems, data assimilation, hemodynamics, air quality.