The implicit Euler method for the numerical solution of singular initial value problems (Q1567643)

The authors study the solvability of systems of singular nonlinear initial-value problems of the form \[ z'(t)= {M(t)\over t} z(t)+ f(t, z(t)),\quad t\in [0,1],\quad B_0z(0)= \beta,\tag{1} \] where \(M\in \mathbb{R}^{n\times n}\) is smooth and \(B_0\in \mathbb{R}^{m\times n}\) \((m\leq n)\). They first discuss under which conditions on the structure of the initial condition bounded solutions exist. The implicit Euler method applied to (1) is then shown to be stable and to retain its classical convergence order, and its classical properties also carry over to the asymptotic expansion of the global error. Numerical examples are used to establish the feasibility of the implicit Euler method as the basic method for the iterated defect correction (IDeC) acceleration method. Two numerical examples illustrate the performance of this IDeC method for singular problems of the type (1).