Error formulas for divided difference expansions and numerical differentiation

Here, and throughout the paper, we will assume that x0 ≤ x1 ≤ . . . ≤ xn are arbitrarily spaced real values and x is any real value in the interval [x0, xn]. We refer the reader to Conte and de Boor [1 ] for basic properties of divided differences. Two things are required: evaluation of the coefficients ck; and a bound on the remainder term Rp in terms of the maximum grid spacing h := max 0≤i≤n−1 (xi+1 − xi).