Probabilistic specifications are fast gaining ground as a tool for statistical modeling of probabilistic systems. One of the main goals of formal methods in this domain is to ensure that specific behavior is present or absent in the system, up to a certain confidence threshold, regardless of the way it operates amid uncertain information. This paper presents a rewriting logic semantics for a probabilistic extension of Event-B, a proof-based formal method for discrete systems modeling. The proposed semantics adequately captures the three sources of probabilistic behavior, namely, probabilistic assignments, parameters, and concurrency. Hence, simulation and probabilistic temporal verification become automatically available for probabilistic Event-B models. The approach takes as input a probabilistic Event-B specification, and outputs a probabilistic rewrite theory that is fully executable in PMaude and can be statistically tested against quantitative metrics. The approach is illustrated with examples in the paper.
翻译:概率性规格作为概率系统统计模型的工具,正在迅速获得立足点。这一领域的正式方法的主要目标之一是确保特定行为在系统中存在或不存在,直至某种信任阈值,而不论在不确定信息中如何运作。本文为不同系统模型的以证据为基础的正式方法,即事件B的概率扩展提供了一个重写逻辑语义,这是独立系统模型的一种基于证据的正式方法。拟议的语义充分捕捉了概率行为的三个来源,即概率性任务、参数和共通货币。因此,模拟和概率性时间核查可以自动用于概率性事件B模型。该方法将概率性事件B规格作为输入,输出出一种在PMaud完全可以执行的概率性重写理论,并可用定量指标进行统计测试。该方法在文件中以实例说明。