IEEE浮点运算下三角函数的表驱动方法

Table-driven Implementation of the Trigonometric Functions for IEEE Floating Points

提出了在ＩＥＥＥ点运算下四倍精度三角函数的算法，特别是“精确”的 三角函数区间压缩。用工作精度（四倍精度）算术运算及适中的表存储使三角函数精确 到 ２．０ｕｌｐ。分析了子程序的误差。大量测试表明函数的精度和执行速度是令人满意的， 并保持了函数的单调性。

Algorithms and implementation details for the trigonometric functions in quad-precision of IEEE 754 arithmetics arc presented. Accurate range reduction with good performance for trigonometric functions is imple mented. With a table of moderate size, the implementation needs only working-precision(quad-precision) arithmetics and are provably accurate to within 2.0 ulp. The error analysis for the subroutines are considered in detail. The testing results show that both the accuracy and the performance are satisfactory. The analytic properties such as monotonicity and others are preserved as well.