In this paper, we study discrete evacuation in networks, where agents know the network topology and designated exit nodes but do not know the number and initial positions of other agents. Each agent initially occupies a distinct node and must reach any exit node. Operating in a synchronous distributed model with local communication, the agents aim to minimize the time when the last agent reaches an exit. We introduce a general algorithmic framework for constructing evacuation strategies on arbitrary graphs. As a key application, we demonstrate that the framework yields asymptotically optimal evacuation strategies -- achieving a constant competitive ratio -- for grid networks, with natural extensions to triangular and hexagonal grids.

翻译：本文研究网络中的离散疏散问题，其中智能体已知网络拓扑结构和指定出口节点，但未知其他智能体的数量和初始位置。每个智能体初始占据不同节点，必须到达任意出口节点。在具有本地通信的同步分布式模型中，智能体旨在最小化最后一个智能体到达出口的时间。我们提出了一种通用算法框架，用于在任意图上构建疏散策略。作为关键应用，我们证明该框架能够为网格网络生成渐近最优的疏散策略——实现恒定竞争比——并可自然扩展至三角形网格和六边形网格。