Starting algorithms for the iterations of the RKN-Gauss methods (Q1807668)

Gauss methods are symplectic implicit Runge-Kutta methods of maximal order. This article considers the numerical solution of the nonlinear Runge-Kutta system, when the method is applied to problems of the form \(y''=f(y)\). Using the information of the previous step, and eventually one or two additional function evaluations, formulas for starting approximations for the fixed point iteration are derived. Some numerical experiments compare the efficiency of different starting procedures.