求解稳定一致的一个充分认知条件
A Sufficient Epistemic Condition for Solving Stabilizing Agreement
AI总结:
论文首次用时态认知逻辑和 Runs and Systems 框架形式化非终止的稳定一致问题,并给出求解该问题的充分认知条件,为后续充要刻画奠定基础。
摘要:
本文给出了有史以来首个稳定一致的认知形式化表述,稳定一致被定义为成熟的共识问题的一种非终止变体。在稳定一致中,各智能体被赋予(可能不同的)初始值,目标是最终始终决定同一个值。虽然允许智能体有限次地改变其决定,但要求它们最终就同一个值达成一致。我们在时态认知逻辑中刻画这些性质,并使用 Runs and Systems(运行与系统)框架对稳定一致问题进行形式化推理。随后,我们从认知上形式化求解稳定一致的条件,并识别在我们的系统假设下,智能体在任意执行过程中为选择某个特定值而获得的知识。这一关于求解稳定一致充分条件的首次形式化,为计划中对稳定一致进行充要的认知刻画奠定了基础。
英文摘要:
In this paper we provide a first-ever epistemic formulation of stabilizing agreement, defined as the non-terminating variant of the well established consensus problem. In stabilizing agreements, agents are given (possibly different) initial values, with the goal to eventually always decide on the same value. While agents are allowed to change their decisions finitely often, they are required to agree on the same value eventually. We capture these properties in temporal epistemic logic and we use the Runs and Systems framework to formally reason about stabilizing agreement problems. We then epistemically formalize the conditions for solving stabilizing agreement, and identify the knowledge that the agents acquire during any execution to choose a particular value under our system assumptions. This first formalization of a sufficient condition for solving stabilizing agreement sets the stage for a planned necessary and sufficient epistemic characterization of stabilizing agreement.
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。