Shape preserving interpolating cubic splines with geometric mesh (Q1192510)

The authors give sufficient conditions for the existence of a cubic \(C^ 2\)-spline, which interpolates in the knots \((x_ i)\) prescribed monotone and convex data. They build on earlier work by \textit{J. W. Schmidt} [Numerical methods of approximation theory, Vol. 8, Workshop Oberwolfach/FRG 1986, ISNM 81, 213-228 (1987; Zbl 0623.41008)] who had presented a recursive algorithm for the construction of such splines, which can, however, become numerically instable in some cases. The approach in the paper under consideration is claimed to cause no numerical instabilities; it is, however, restricted to so-called geometric meshes. This means that the knot sequence \((x_ i)\) satisfies \({{x_{i+2}-x_{i+1}} \over {x_{i+1}-x_ i}}=q=\)constant.