Digital computers cannot, in general, evaluate values with infinite precision and thus need to rely on approximations. It is, of course, among the goals of any scheme resulting in such an approximate value to have as little an error as possible. Numerical schemes to reduce this error, such as the Richardson extrapolation presented here, are therefore of great interest.
The Richardson extrapolation scheme assumes that a function can be expanded around a point of evaluation depending on a parameter in a form
Multiplying Equation by and subtracting Equation allows to eliminate the terms of order
Applications using digital computers usually utilize factors corresponding to powers of . This allows the following settings
which further illustrate the recursive nature of the procedure.