We propose a two-level information-theoretic framework for characterizing the informational organization of Agent-Based Model (ABM) dynamics within the broader paradigm of Complex Adaptive Systems (CAS). At the macro level, a pooled $\epsilon$-machine is reconstructed as a reference model that summarizes the system-wide informational regime. At the micro level, $\epsilon$-machines are reconstructed for each caregiver-elder dyad and variable, and are complemented with algorithm-agnostic Kolmogorov-style measures, including normalized LZ78 complexity and bits per symbol from lossless compression. The resulting feature set $\{h_{\mu}, C_{\mu}, E, \mathrm{LZ78}, \mathrm{bps}\}$ enables distributional analysis, stratified comparisons, and unsupervised clustering across agents and scenarios. This dual-scale design preserves agent heterogeneity while providing an interpretable macro-level baseline, aligning ABM practice with CAS principles of emergence, feedback, and adaptation. A case study on caregiver-elder interactions illustrates the framework's implementation; the results and discussion will be completed following final simulation runs.

翻译：本文提出一种双层信息论框架，用于在复杂适应系统（CAS）的宏观范式下刻画基于智能体的模型（ABM）动态的信息组织结构。在宏观层面，重构聚合ε-机作为参考模型，以概括系统整体的信息状态。在微观层面，为每个护理者-老人配对及变量重构ε-机，并辅以与算法无关的柯尔莫哥洛夫式度量，包括归一化LZ78复杂度与无损压缩的每符号比特数。所得特征集 $\{h_{\mu}, C_{\mu}, E, \mathrm{LZ78}, \mathrm{bps}\}$ 支持跨智能体与场景的分布分析、分层比较及无监督聚类。该双尺度设计在保持智能体异质性的同时，提供了可解释的宏观基准，使ABM实践与CAS的涌现、反馈及适应原则相统一。以护理者-老人互动为例的案例研究展示了该框架的实施；最终仿真运行完成后将补充结果与讨论部分。