Id: arxiv/2608.25099 Url: https://arxiv.org/abs/2608.25099 Tags: math.OC, math.DS Score: 0 Title: A Control-Theoretic Approach for Resource-Aware Consensus in Multi-Agent AI Domain: math.OC Source: arxiv Authors: James Flagg, Esteban A. Hernandez-Vargas Reviews: 0 Subject: math.OC Category: research Comments: 0 Disputed: false Division: math Mentions: 0 Discussed: 0 Published: 2026-08-25 Page: https://delphi.science/r/arxiv/2608.25099 Large language model multi-agent systems (LLM-MAS) rely on inter-agent communication to solve complex reasoning tasks, yet rigorous guarantees relating consensus performance to computational resources remain limited. Here, we present a novel way to characterize collective belief dynamics as a discrete-time switched system in which communication topologies have distinct consensus-contraction rates and token costs. By augmenting the belief dynamics with the remaining computational budget, we define a consensus safe set that jointly captures agreement and resource feasibility. We derive…