对于有吸收目标状态的循环任务，比较合理的方法是采用基于平均报酬模型的强化学习。

It is rational to adopt the average reward reinforcement learning algorithms for solving the absorbing goal states cyclical tasks.

youdao

花费大部分时间在库函数调用上的代码不会像那些简单循环任务一样使用Pyrex可以获得明显的加速。

Code that spends most of its time making library calls is just not going to benefit that much from Pyrex speeding up incidental loop overhead.

youdao

该算法通过识别并行循环任务以及通过确定调用上下文关系识别并行过程任务，来划分串行程序中的并行成份。该算法是对串行程序实施并行分解的基础。

This algorithm is the basis for decomposing parallelism in a sequential program and is implemented with recognizing parallel loop tasks and parallel procedural tasks by determining the call context.

youdao