The numerical solution of nonlinear differential equations by spline functions (Q1207384)

A method to construct spline functions which approximate the solution \(y: [0,1]\to\mathbb{R}\) of a nonlinear initial value problem for an \(m\)th order differential equation is proposed. Assuming that the differential equation is sufficiently smooth, the authors suggest to use a Taylor series to obtain approximations to the solution and its derivatives up to the second order at the points of a uniform grid in \([0,1]\). Then by using a procedure employed by \textit{T. Fawzy} [Acta Math. Acad. Sci. Hungar. 29, 259-271 (1977; Zbl 0448.65049)] they construct a spline function in the interval \([0,1]\) which is the desired approximation to the solution. Finally, it is proved that for small values of \(m\) (the degree of the differential equation), the spline function converges to the solution of the differential equation.