Generation of interpolation curves with the least energy using dynamic programming (Q2365965)

This paper is devoted to the classical interpolation problem in one dimension, but with an additional condition, namely, that the interpolating curve \(y(\cdot)\) has minimal ``strain energy'' in the sense of the integral \(\int^{x_{n-1}}_{x_ 0}[(y'')^ 2/[1+(y')^ 2]^{5/2}]dx\), i.e. \(y(\cdot)\) satisfies the usual interpolation conditions \(y(x_ i)=y_ i\) for \(i=0,\dots,n-1\), plus \(y'(x_ i)=d_ i\) computed by an ``arbitrary method'' (?) and gives the minimum of the above integral. Unfortunately, by a lot of essential mistakes, this paper is unintelligible and, in my opinion, is false from mathematical point of view. The dynamic programming method is used for determination of the optimal interpolating curve.