计算机浮点数算术运算的舍入误差研究  

Rounding error study of calculator floating-point numbers arithmetic operation

作　　者：范啸涛[1] 季光明[1] 何永斌[2] 

机构地区：[1]成都理工大学信息管理学院,成都610059 [2]温州大学(筹)计算机科学与工程学院,温州325035

出　　处：《成都理工大学学报（自然科学版）》2005年第2期213-216,共4页Journal of Chengdu University of Technology: Science & Technology Edition

摘　　要：对计算机浮点数算术运算的舍入误差进行分析,是对数值计算方法作误差分析的基础。论文全面研究了计算机浮点数算术运算的舍入误差的基本理论,对有关结论均作了严格的论证。并举例说明由算术运算的误差界,可建立较复杂运算的误差界。还研究了向前误差分析和向后误差分析的要点。推广上述结果,对矩阵基本运算的舍入误差进行了估计。To analyze the rounding error of calculator floating-point numbers arithmetic operation is the foundation of numerical calculation method’s error analysis. This paper studies roundly the basic theories of rounding error of calculator floating-point numbers arithmetic operation and demonstrates the related conclusions. It also illustrates with examples that the error margin calculated by arithmetic operation can establish more complicated calculation error margin.This paper still studies the mains of the forward error analysis and backward error analysis. By generalizing the above results, the paper estimates the rounding error of matrix basic calculation.

关 键 词：浮点数系统 舍入误差 机器精度 向后误差分析 

分 类 号：O242.2[理学—计算数学]