Estimation of the global discretization error in shooting methods for linear boundary value problems

摘要

In this paper we consider an error estimation procedure in the numerical solution of a linear boundary value problem for a system of m first-order equations when the simple or parallel shooting method is used. We show that the global discretization error can be estimated through the numerical solution of the principal error equation related to only one initial-value problem, rather than the m + 1 involved by shooting. Four significant numerical experiments are presented.