# split in half m = n / 2 # recursive sorts sort a[1..m] sort a[m+1..n] # merge sorted sub-arrays using temp array b = copy of a[1..m] i = 1, j = m+1, k = 1 while i <= m and j <= n, a[k++] = (a[j] < b[i]) ? a[j++] : b[i++] → invariant: a[1..k] in final position while i <= m, a[k++] = b[i++] → invariant: a[1..k] in final position
Merge sort is very predictable. It makes between 0.5*lg(n) and lg(n) comparisons per element, and between lg(n) and 1.5*lg(n) swaps per element. The minima are achieved for already sorted data; the maxima are achieved, on average, for random data. If using Θ(n) extra space is of no concern, then merge sort is an excellent choice: It is simple to implement, and it is the only stable O(n·lg(n)) sorting algorithm. Note that when sorting linked lists, merge sort requires only Θ(lg(n)) extra space (for recursion).
Merge sort is the algorithm of choice for a variety of situations: when stability is required, when sorting linked lists, and when random access is much more expensive than sequential access (for example, external sorting on tape).
There do exist linear time in-place merge algorithms for the last step of the algorithm, but they are both expensive and complex. The complexity is justified for applications such as external sorting when Θ(n) extra space is not available.
Bubble-sort with Hungarian ("Csángó") folk dance YouTube video, created at Sapientia University, Tirgu Mures (Marosvásárhely), Romania.
Select-sort with Gypsy folk dance YouTube video, created at Sapientia University, Tirgu Mures (Marosvásárhely), Romania.
Sorting Out Sorting, Ronald M. Baecker with the assistance of David Sherman, 30 minute color sound film, Dynamic Graphics Project, University of Toronto, 1981. Excerpted and reprinted in SIGGRAPH Video Review 7, 1983. Distributed by Morgan Kaufmann, Publishers. Excerpt.