This paper studies the basins of attraction of fully asynchronous cellular automata (ACAs) that converge to point attractor. Unlike traditional (synchronous) cellular automata (CAs), an ACA may approach to different point attractors from the same initial condition for different runs. In this direction, this paper introduces the notion of cloud behavior in convergence dynamics of ACAs. Depending on the cloud behavior, we classify the convergent ACAs into three classes—eccentric cloud, partially eccentric cloud and deterministic cloud system. In addition, this study observes the computational ability of addressing density classification problem (i.e. distributed consensus problem) of this asynchronous cloud system.

本文研究了收敛到点吸引子的完全异步元胞自动机 (ACA) 的吸引力盆地。与传统（同步）元胞自动机 (CA) 不同，ACA 可能会针对不同的运行从相同的初始条件接近不同的点吸引子。在这个方向上，本文介绍了 ACA 收敛动力学中云行为的概念。根据云的行为，我们将收敛的 ACA 分为三类——偏心云、部分偏心云和确定性云系统。此外，本研究还观察了该异步云系统解决密度分类问题（即分布式共识问题）的计算能力。