Keywords
The general formula for an
Moreover, for
There are two popular mathematical formats to express this polynomial: the Newton and the Lagrange polynomials.
Newton’s Divided-Difference Interpolating Polynomials
Linear Interpolation
Linear interpolation, the simplest form of interpolation, involves connecting two data points using a straight line.
Characteristics
- First order interpolating polynomial.
- The accuracy of its approximation increases as the interval between the data points decreases.
Formula
The term
Quadratic Interpolation
We can obtain a second-order interpolating polynomial (or a quadratic polynomial) when we have three data points given. A convenient for this purpose is
where
The term
General Form
The general form for an
For an
To find the
Graphical Depiction of the Nature of Finite Divided Difference
First Second Third
Errors
The
where
The issue with higher-order interpolating polynomials, however, is that they are highly sensitive to data errors. When interpolation is applied, they often yield inaccurate predictions. For this reason, they are better suited for exploratory data analysis.
Lagrange Interpolating Polynomials
The Lagrange interpolating polynomial is a reformulation of the Newton’s polynomial that avoids divided differences, thus
where
where
The error estimate can be computed with
Newton or Lagrange?
For exploratory computations, Newton’s method is preferred because it gives insight to the different-order formulas’ behavior; it is suited for cases where the polynomial’s order is unknown. Furthermore, because Newton’s method employs a finite difference, we can easily integrate the error estimate to it.
Lagrange is preferred over Newton in cases where only one interpolation is to be performed because it is easier to program—it does not depend on divided differences, but the polynomial’s order must be known beforehand.
Sources
- Numerical Methods for Engineers by Steven Chapra and Raymond Canale (Chapter 18)