A quantitative understanding of the heart function is crucial for several cardiac therapies, ranging from ablation or resynchronization therapy to valve replacement. The electrical myocard excitation is described by the mono- or bidomain equations, a set of reaction-diffusion equations for the transmembrane voltage and a set of pointwise ODEs for the ion concentrations and ion channels. Due to very different temporal and spatial scales involved, the numerical simulation is rather expensive. In this project we develop different structure-exploiting integration methods for mono- and bidomain equations, in order to reduce the computational effort while controlling the discretization error.
Multilevel Methods for Scar Detection
In close collaboration with USI, we investigate the use of heterogeneous multilevel methods consisting of monodomain fine model and eikonal coarse model within a recursive multigrid trust region approach for identifying myocardial scars from endocardial mapping data.MICROCARD
Since 2021, we participate in the MICROCARD project funded by the EU Joint Undertaking HPC and BMBF. The project aims at developing numerical software able to simulate cardiac excitation with a cellular resolution at organ scale on advanced HPC exascale systems.
Within MICROCARD, we work on energy-efficient time discretization schemes, hierarchical and lateral model coupling for multilevel solution and boundary conditions, respectively, and compressed communication in domain decomposition preconditioners.
