New algorithms for the continuous approximate solution of ordinary differential equations and the upgrading of the order of the processes (Q915389)

For solving ordinary differential equations numerically there exists a wide range of methods. Embedding a type six stage fifth order Runge-Kutta formula yields a continuous approximation of a first order ordinary differential equation. The paper first presents three methods of this kind. Then a second order ordinary differential equation is solved by first transforming it to a system of first order. Above methods then furnish continuous approximations of the solution y(t) and its derivative \(y'(t)\). Integrating the approximations of \(y'(t)\) gives approximations of y(t). The author shows that they approximate the exact solution of one order higher than the corresponding original approximations to y(t) provided by Runge-Kutta. A similar upgrading technique can be applied to higher order ordinary differential equations and implicit differential equations of the form \(F(x,y,y',y'',...,y^{(n)})=0\), where the highest derivative cannot be written in explicit form. Such an upgrading, however, is not possible with conventional discrete methods. The author provides error estimations for his techniques. A large number of detailed numerical examples concludes the paper.