Analysis of parallel diagonally implicit iteration of Runge-Kutta methods (Q686552)

The main aim of the paper is to compare the parallel diagonally implicit iteration of Runge-Kutta methods (PDIRK) earlier developed by the authors [SIAM J. Sci. Stat. Comput. 12, No. 5, 1000-1028 (1991; Zbl 0732.65065) and Math. Comput. 58, No. 197, 135-159 (1992; Zbl 0744.65050)] with the methods based on backward differentiation (BDF), considered as one of the best for sequential computers. For large systems of stiff equations, the PDIRK methods are comparable to the Newton iteration BDF's, but they have the advantage that they can be parallelized. Another advantage of the authors' methods is that they are \(A\)-stable and in some cases even \(L\)-stable. Special characteristics of the PDIRK methods are studied such as: the rate of convergence and the influence of predictors on stability. The paper concludes with some practical recommendations concerning the PDIRK methods.