Efficient block predictor-corrector methods with a small number of corrections (Q688031)

For the numerical solution of nonstiff ordinary differential equations in a parallel environment block predictor-corrector methods are considered. A numerical example illustrates that the application of low-order predictors is not very efficient. Consequently, the use of high-order predictors together with a few correction steps is proposed. Based on a detailed error analysis of such methods (using \(B\)-series) a new corrector (collocation method) is derived.