Multi agent simulations maintain diversity with prospect state modeling
Scaling Multi-Agent Systems with Prospect-State Propagation
Multiagent SystemsComputers and Society
Summary
Large groups of AI agents sometimes lose their unique behaviors over time, making simulations less realistic. The authors noticed this problem in economic simulations and created a new way to keep agents different from each other. They split each agent’s state into two parts: a simple part that keeps track of psychological differences, and a complex part handled by language models. This method helps keep the system diverse and scalable without losing important details.
What this means in practice
- •For economic modelers: Create large-scale economic simulations that preserve diverse agent behaviors for more realistic macroeconomic forecasting.
- •For game developers: Design multi-agent AI with sustained diverse behaviors to enhance realism and complexity in simulation-based games.
Authors
Zhimei Chen, Mu Chen, Fakhri Karray
Abstract
Current LLM-based multi-agent systems (MAS) periodically compress intermediate states to reduce inference-time token consumption, thereby attempting to incorporate more agents. However, naive scaling strategies face challenges. For example, in economic simulations, large-scale MAS typically discard semantically rich economic states, i.e., agent behavioral trajectories, which are key drivers of macroeconomic fluctuations. In this paper, we reveal a phenomenon in which agent heterogeneity gradually decreases during simulation, and propose Prospect-State Propagation for Multi-Agent Systems (PspMAS). Inspired by prospect theory, PspMAS decouples each agent's micro state into a compact Prospect State and an expressive Semantic State. The former records psychological traces through a lightweight, parallelizable propagator and continuously injects heterogeneity into the system. The latter leverages the strong perception, reasoning, planning, and decision-making abilities of LLMs. These two components work complementarily, providing a scalable LLM-based multi-agent simulation solution.