科学计算程序语言的浮点数机制研究

Research on Mechanism of Floating-point Value for Program Language in Science Computation

浮点数运算存在精度方面、比较方面以及舍入误差等方面的问题,而这些问题直接影响到科学计算的准确性、可靠性和安全性等等。目前有关浮点数的中文资料很少,很多教科书上在谈到浮点数时都是浅尝辄止,本文以C语言浮点数机制为研究基础,对浮点数的格式、精度与应用等方面问题进行了实证研究,获得了一些有用的结果。

There exist some problems in floating-point operation, such as problems of accuracy, comparison, rounding error, and so on. They have direct affects to accuracy, reliability and security of scientific computing etc. At present, there are only a few Chinese documents about floating-point value. It scratches the surface of floating-point value in many textbooks. The paper carries out a demonstration research on format, accuracy and application based on floatingpoint mechanism of C language and some useful results have been obtained.