Global and local data reduction strategies for cubic splines (Q756911)

The author proposes two algorithms which allow the user to minimize the number of knots of an approximating cubic spline s. The cubic spline is forced to satisfy \(y_ i-a_ i\leq s(x_ i)\leq y_ i+b_ i\quad 1\leq i\leq n,\) where \((x_ i,y_ i)\) are the given points and \((a_ i,b_ i)\) are assumed nonnegative tolerances. Numerical examples are included.