A new \(C^2\) rational interpolation based on function values and constrained control of the interpolant curves (Q1763243)

Constructed the rational interpolation spline for approximation function from \(C^2[a,b].\) Let \(\{(t_i, f_i),\) \(i=0,1,\dots,n,n+1\}\) be a given set of data points, where \(a=t_0 < t_1 < \dots < t_n < t_{n+1}=b\) is the knot spacing. The \(C^1 -\)continuous, piecewise, rational cubic function based on the function values is defined by \[ P(t)|_{[t_i,t_{i+1}]}=\frac{p_i(t)}{q_i(t)}, \quad i=0,1,\dots,n-1, \tag{1} \] where \[ \begin{aligned} p_i(t)&= (1-\Theta)^3 \alpha_i f_i + \Theta(1-\Theta)^2V_1 + \Theta^2(1-\Theta)W_i + \Theta^3 \beta_i f_{i+1},\\ q_i(t)&= (1-\Theta)^2 \alpha_i + 2\Theta(1-\Theta)+\Theta^2\beta_i,\\ \Theta&=(t-t_i)/h_i,\\ h_i&= t_{i+1}-t_i,\\ V_i&=2f_i+\alpha_i f_{i+1}, \\ W_i&=(2+\beta_i)f_{i+1} - h_i \beta_i \Delta_{i+1}, \end{aligned} \] with \(\alpha_i, \beta_i > 0,\) and \(\Delta_i = (f_{i+1}-f_i)/h_i.\) It is shown that one can choose parameters \(\alpha_i\) and \(\beta_i\) in such a way, that the rational interpolation spline \(P(y) \in C^2[a,b]\) and the diagram of \(P(y)\) lays between two piece-linear functions, which are determined before. Exactness of the method is estimated. There are given numerical examples.