A General Lower Bound on the I/O-Complexity of Comparison-based Algorithms

A General Lower Bound on the I/O-Complexity of Comparison-based Algorithms
复制标题

基于比较的算法的 I/O 复杂度的一般下界

DOI:
--
复制
发表时间:
1992
期刊:
Workshop on Algorithms and Data Structures
影响因子:
--
通讯作者:
Kirsten Larsen
Kirsten Larsen
中科院分区:
--
文献类型:
--
作者:
L. Arge;Mikael B. Knudsen;Kirsten Larsen

文献摘要

被引文献

相似文献

我们展示了解决给定问题所需的比较次数和I/O操作次数之间的一般关系。这种关系使人们能够显示下限的数量的I/O操作需要解决一个问题时,一个较低的边界上的数量的比较是已知的。我们使用的结果,显示下界的I/O复杂性的一些问题,已知的技术只给平凡的界限。其中包括从一个多重集合中删除重复项的问题,这是一个在关系数据库系统中非常重要的问题,以及确定一个多重集合的模式(最频繁出现的元素)的问题。我们开发这些问题的算法,以表明,下界是紧的。
We show a general relationship between the number of comparisons and the number of I/O-operations needed to solve a given problem. This relationship enables one to show lower bounds on the number of I/O-operations needed to solve a problem whenever a lower bound on the number of comparisons is known. We use the result to show lower bounds on the I/O-complexity on a number of problems where known techniques only give trivial bounds. Among these are the problems of removing duplicates from a multiset, a problem of great importance in e.g. relational data-base systems, and the problem of determining the mode — the most frequently occurring element — of a multiset. We develop algorithms for these problems in order to show that the lower bounds are tight.