A Fast Sorting Algorithm, a Hybrid of Distributive and Merge Sorting

Distributive sort methods split the input data into smaller groups Gi , . . . , G,, such that records in group Gi will have smaller keys than records in group Gj if i < j. Each of these groups is split into even smaller groups in the same way. The process is continued until sorted or single-element subgroups remain. In most cases it pays to sort very small groups by a more traditional method, such as straight insertion sort. Some authors (e.g. [8]) distinguish two types of distributive sorting: radix and range sorti:ig. In the former partitioning is based on the value of the key as a whole. The latter makes use of specific digits. Recently, Dobosiewicz introduced a new class of range sorts [2]. Its principle is very simple: if there are n records to be partitioned into m groups, then a record with key value (x belongs in group Gj with: 1