Séminaire de Stéphanie Chaillat

Boundary Element Methods (BEM), based on the discretization of boundary integral equations, have proven particularly well suited for modeling waves in unbounded domains. In recent years, significant advances have been achieved within the BEM community to make these methods applicable to realistic configurations, notably through the development of fast techniques (fast multipole method, low-rank approximations, mesh adaptation) that overcome the memory constraints inherent to the method, particularly the dense nature of the system matrix.

Today, fast BEM have reached maturity and play a key role in the simulation of multi-physics phenomena, especially when coupled with other numerical methods such as FEM. This talk will present the fundamental principles of fast BEM, their connection with big data and machine learning, as well as coupling strategies with other numerical approaches to solve complex problems. The capabilities of these methods will be illustrated through various applications, including industrial ones, such as seismic wave propagation and underwater acoustics.