Efficient unbalanced merge-sort

被引：4
|
作者：
Nardelli, E
Proietti, G
机构：
[1] Univ Aquila, Dipartimento Informat, I-67010 Coppito, Italy
[2] Univ Roma Tor Vergata, Dipartimento Matemat, I-00133 Rome, Italy
[3] CNR, Ist Anal Sistemi & Informat A Ruberti, I-00185 Rome, Italy
来源：
INFORMATION SCIENCES | 2006年 / 176卷 / 10期
关键词：
design of algorithms; sorting; experimental analysis; data structures;
D O I：
10.1016/j.ins.2005.04.008
中图分类号：
TP [自动化技术、计算机技术];
学科分类号：
0812 ;
摘要：
Sorting algorithms based on successive merging of ordered subsequences are widely used, due to their efficiency and to their intrinsically parallelizable structure. Among them, the merge-sort algorithm emerges indisputably as the most prominent method. In this paper we present a variant of merge-sort that proceeds through arbitrary merges between pairs of quasi-ordered Subsequences, no matter which their size may be. We provide a detailed analysis.. showing that a set of n elements can be sorted by performing at most n[logn] key comparisons. Our method has the same optimal asymptotic time and space complexity as compared to previous known unbalanced merge-sort algorithms, but experimental results show that it behaves significantly better in practice. (c) 2005 Elsevier Inc. All rights reserved.
引用
收藏
页码：1321 / 1337
页数：17
相关论文
共 50 条
12345