Automatic solution of optimal-control problems. I. Simplest problem in the calculus of variations (Q760176)

The authors propose a ''table method'' for ''automatic'' (i.e. by computer) calculation of derivatives of functions of several variables that may be obtained from (elementary) functions of one variable via composition and the main algebraic operations: addition, subtraction, multiplication and division. The table method is a modification of \textit{R. Wengert}'s method [Commun. ACM 7, 463-464 (1964; Zbl 0131.346] and is used to approximate by the Newton-Raphson algorithm the solution of the two point boundary- value Euler-Lagrange equation associated with the problem of minimizing \(\int^{b}_{a}F(t,x(t),x'(t))dt\) subject to: \(x(a)=c\), \(x(b)=d\), defined by F(.,.,.): \(R^ 3\to R\). Numerical experiments for \(F(t,x,x')=(1+(x')^ 2)^{1/2}/x,\) \(x(1)=1\), \(x(2)=2\) and a FORTRAN program listing are presented. The authors state that ''the user of the program need only input the integrand... and specify the boundary conditions'' but it seems better that prior to that, the user should ask himself (or a mathematician) whether the minimization problem has a solution, whether the Euler-Lagrange B.V.P. has a unique solution, etc., for what happens if, for instance \(F(t,x,x')=(x^ 2+(x')^ 2)^{1/2}\), \(x(0)=0\), \(x(1)=1\) [cf. \textit{L. Cesari}, Optimization-theory and applications. Problems with ordinary differential equations (1983; Zbl 0506.49001); Example 5, {\S} I.5] ?