基于多核CPU的表约束并行传播模式研究

并行传播是并行约束程序领域中的一个研究方向，其研究内容是如何并行执行在约束上的过滤算法.根据维持表约束网络广义弧相容（generalized arc consistency，简称GAC）的串行传播模式，提出了维持表约束网络临时广义弧相容（temporary generalized arc consistency，简称TGAC）的并行传播模式，该模式基于多核CPU，由并行传播算法和并行过滤算法两部分组成；之后，给出了并行传播模式的可靠性证明，而且通过对并行传播模式的最坏时间复杂度分析，可以认为并行传播模式在平均过滤时间较长的实例上要快于串行传播模式；最终的实验结果也验证了上述结论，并行传播模式在多数实例集上取得了从1.4~3.4不等的加速比.

Research on Parallel Propagation Mode of Table Constraint Based on Multi-core CPU

Parallel propagation is a research direction in the field of parallel constraint programming, and its research content is how to implement filtering algorithms on constraints in parallel. According to the serial propagation mode which enforces generalized arc consistency (GAC) on table constraint network, this study proposes a parallel propagation mode to enforce temporary generalized arc consistency (TGAC) on table constraint network. This mode is based on multi-core CPU and consists of parallel propagation algorithm and parallel filtering algorithm. After that, the reliability of the parallel propagation mode is proved, and through the analysis of the worst case time complexity of the parallel propagation mode, it is also demonstrated that the parallel propagation mode is faster than the serial propagation mode in instances of which the average filtering time is longer. Finally, the experimental results also confirm the above conclusion, and the parallel propagation mode achieves a speed-up ratio ranging from 1.4 to 3.4 on most series.