A new numerical method by revised measure theory for solving the nonlinear initial value problems (Q876661)

This paper is concerned with the numerical solution of initial value problems (IVPs) for \(n\)th order differential equations \[ y^{(n)} = F(t, y, y', \ldots , y^{(n-1)}), \quad y^{(j)}(t_a) = y^{(j)}_a, j=0, \ldots,n-1,\;t \in [t_a, t_b].\tag{1} \] After transforming this equation into an equivalent first order system in the standard form \( x'(t)= g(t, x(t)), x(t_a)= x_a \in \mathbb{R}^n\), \( t \in [t_a, t_b] (1)\), the authors start a mathematical metamorphosis of (1) with several stages. In the first stage, by introducing an artificial cost function, the IVP (1) is written as a minimization problem over a set of Radon measures. Then, in a second stage, the problem is approximated by a finite dimensional linear programming problem. It is unclear what are the advantages of this approach for solving IVPs because the original IVP (1) is transformed into a more complicated problem and some approximations are also introduced, and no results on the accuracy and stability properties of this method are given. Finally, some numerical experiments with a linear second order equation and a first order equation are presented.