1: \begin{abstract}

2: We present a novel approach which aims at high-performance uncertainty

3: quantification for cardiac electrophysiology simulations. Employing

4: the monodomain equation to model the transmembrane potential inside

5: the cardiac cells, we evaluate the effect of spatially correlated

6: perturbations of the heart fibers on the statistics of the resulting

7: quantities of interest. Our methodology relies on a close integration of

8: multilevel quadrature methods, parallel iterative solvers

9: and space-time finite element discretizations, allowing for a fully

10: parallelized framework in space, time and stochastics.

11: Extensive numerical studies are presented to

12: evaluate convergence rates and to compare the performance of

13: classical Monte Carlo methods such as standard Monte Carlo (MC)

14: and quasi-Monte Carlo (QMC), as well as multilevel strategies,

15: i.e.\ multilevel Monte Carlo (MLMC) and multilevel

16: quasi-Monte Carlo (MLQMC) on hierarchies of nested meshes.

17: We especially also employ a recently suggested variant

18: of the multilevel approach for non-nested meshes

19: to deal with a realistic heart geometry.

20: \end{abstract}

21: