Consider an arbitrary sequence
comprised of of elements drawn from a some universal set *U*.
The goal of *sorting* is to rearrange the elements of *S*
to produce a new sequence, say *S*',
in which the elements of *S* appear *in order*.

But what does it mean for the elements of *S*' to be *in order*?
We shall assume that there is a relation, *<*,
defined over the universe *U*.
The relation *<* must be a *total order*,
which is defined as follows:

DefinitionAtotal orderis a relation, say<, defined on the elements of some universal setUwith the following properties:

In order to *sort* the elements of the sequence *S*,
we determine the *permutation*
of the elements of *S* such that

In practice, we are not interested in the permutation *P*, *per se*.
Instead, our objective is to compute the sorted sequence
in which for .

Sometimes the sequence to be sorted, *S*, contains duplicates.
I.e., such that .
In general when a sequence that contains duplicates is sorted,
there is no guarantee that the duplicated elements
retain their relative positions.
I.e., could appear either before or after
in the sorted sequence *S*'.
If duplicates retain their relative positions in the sorted sequence
the sort is said to be *stable* .
In order for and to retain their relative order in the
sorted sequence,
we require that precedes in *S*'.
Therefore, the sort is stable if .

Copyright © 1997 by Bruno R. Preiss, P.Eng. All rights reserved.