High performance parallel numerical methods for Volterra equations with weakly singular kernels (Q1019660)

The authors investigate high performance parallel numerical methods for Volterra equations with weakly singular kernels. They construct fast convergent discrete time nonstationary wave form relaxation (NSWR) methods using a fractional linear multistep formula. They construct the family of Richardson fully parallel discrete time NSWR methods and give the expression of the parameters of the optimal methods with respect to the convergence rate. They provide numerical experiments that confirm the theoretical results and show that the convergence improvement with respect to the optimal stationary methods varies from the 33\% to the 75\%.