A constructive approach to cubic Hermite fractal interpolation function and its constrained aspects (Q2434939)

The authors present a simple explicit construction for a \(C^1\) cubic Hermite fractal interpolation function. Based on the theory of an iterated function system, the concept of fractal interpolation function (FIF) was introduced by \textit{M. F. Barnsley} [Constructive Approximation 2, 303--329 (1986; Zbl 0606.41005)]. Let \(x_1 < x_2 < \ldots < x_N\) be a partition of the interval \(I =[x_1, x_N]\). Let \(y_n \in {\mathbb R}\) be the function value and let \(d_n \in {\mathbb R}\) be the derivative value at the node \(x_n\) \((n=1,\dots,N)\). Then the \(C^1\) cubic Hermite FIF \(s\) fulfils \(s(x_n) = y_n\), \(s'(x_n) = d_n\) \((n=1,\dots,N)\). For \(1\leq p \leq \infty\), the \(L_p(I)\)-errors \(\|(s - h)^{(j)}\|_p\) \((j=0,1)\) between \(s\) and the classical \(C^1\) cubic Hermite interpolant \(h\) are estimated. Later, the derivative values \(d_n\) \((n=1,\dots, N)\) are used as free parameters such that \(s\in C^2\) or the graph of \(s\) lies within a prescribed rectangle. Hence the positivity of \(s\) can be ensured. Numerical examples are given.