For a given sequence
the transformed sequence is
where the members of the transformed sequence are usually computed from some finite number of members of the original sequence, i.e.
In the context of acceleration of convergence, the transformed sequence is said to converge faster than the original sequence if
where is the limit of , assumed to be convergent. In this case, convergence acceleration is obtained. If the original sequence is divergent, the sequence transformation acts as extrapolation method to the antilimit .
If the mapping is linear in each of its arguments, i.e., for
for some constants (which may depend on n), the sequence transformation is called a linear sequence transformation. Sequence transformations that are not linear are called nonlinear sequence transformations.
Simplest examples of (linear) sequence transformations include shifting all elements, (resp. = 0 if n + k < 0) for a fixed k, and scalar multiplication of the sequence.
A little less trivial generalization would be the discrete convolution with a fixed sequence. A particularly basic form is the difference operator, which is convolution with the sequence and is a discrete analog of the derivative. The binomial transform is another linear transformation of a still more general type.
An example of a nonlinear sequence transformation is Aitken's delta-squared process, used to improve the rate of convergence of a slowly convergent sequence. An extended form of this is the Shanks transformation. The Möbius transform is also a nonlinear transformation, only possible for integer sequences.
- Hugh J. Hamilton, "Mertens' Theorem and Sequence Transformations", AMS (1947)