作为能量博弈的进程等价问题
Process Equivalence Problems as Energy Games
AI总结:
本研究提出用6维能量博弈统一刻画所有常见行为等价概念,以防御方获胜初始信用值确定进程间的预序与等价关系,复杂度为最优指数级,在VLTS基准上表现与最优相似性算法相当。
摘要:
我们通过一种6维能量博弈刻画了所有常见的行为等价概念,其中能量约束了试图区分进程的攻击者的能力。防御方获胜的初始信用值可彻底确定(强)线性时间-分支时间谱中的哪些预序和等价关系可关联进程。该算法时间复杂度为指数级,由于覆盖了迹等价,这一复杂度是最优的。这一复杂度相较我们近期判定等价组的方法有大幅提升,后者需在超指数级可达博弈之上构建可区分的HML公式的指数级集合。在使用VLTS基准的实验中,该算法的表现与最优的相似性算法相当。
英文摘要:
We characterize all common notions of behavioral equivalence by one 6-dimensional energy game, where energies bound capabilities of an attacker trying to tell processes apart. The defender-winning initial credits exhaustively determine which preorders and equivalences from the (strong) linear-time--branching-time spectrum relate processes. The time complexity is exponential, which is optimal due to trace equivalence being covered. This complexity improves drastically on our recent approach for deciding groups of equivalences where exponential sets of distinguishing HML formulas are constructed on top of a super-exponential reachability game. In experiments using the VLTS benchmarks, the algorithm performs on par with the best similarity algorithm.
机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。