摘要： 针对归纳变量的识别及归约问题提出一种基于符号运算的新算法。通过初始变量替换赋值语句中的右值，用以表达变量间的迭代与依赖关系，由此建立有向依赖图用于识别归纳变量。为归纳变量构建递归方程组并利用符号运算进行求解，获得独立的仅依赖于迭代次数的数学形式。实验结果表明，该方法适用于各种复杂的归纳变量，能够解决现有算法无法处理的一些问题。

关键词: 静态检测, 程序分析, 代码优化, 符号运算

Abstract: Based on symbolic operation, a novel approach is proposed to recognize and reduce Induction Variables(IV). It replaces all l-values in assignments by the original variables which can represent the iteration and dependence relationships among them, and constructs variable dependence digraphs to recognize IV. It establishes recursive equation sets to solve the problems based on symbolic operations, and gets their independent expressions composed of iteration number. Experimental results show that the method fits with different kinds of IV and overcomes the shortages of current methods.

Key words: static detection, program analysis, code optimization, symbolic operation

中图分类号: 

• TP311