MMAO-Dyn: A Metabolic Multi-Agent Optimizer for Dynamic Optimization
2026-07-01 • Neural and Evolutionary Computing
Neural and Evolutionary Computing
AI summaryⓘ
The authors explore if the Metabolic Multi-Agent Optimizer (MMAO) can be adapted to handle changing environments without adding outside adjustment parts. They created MMAO-Dyn, which adjusts internal processes like energy use and role changes to respond to shifts in problem landscapes. Testing on various dynamic benchmark problems showed MMAO-Dyn performs better than several other dynamic methods, especially in recovering after changes. However, some comparisons were mixed, and improvements mainly come from smarter resource shifts rather than completely redesigning the method. The authors suggest MMAO-Dyn is a promising step in making MMAO work well for dynamic problems.
Metabolic Multi-Agent Optimizerdynamic optimizationnonstationary environmentbenchmark landscapessphere functionAckley functionRastrigin functionresource redistributionalgorithm recoveryagent-based optimization
Authors
Jinliang Xu, Liping Ma
Abstract
This paper studies whether the Metabolic Multi-Agent Optimizer (MMAO) can be credibly derived into a dynamic-optimization method without replacing its core metabolic control loop by external adaptation modules. The proposed MMAO-Dyn maps private energy, communal budget, role drift, success feedback, and lifecycle turnover to a nonstationary setting in which environmental changes repeatedly invalidate previously useful local structure. We evaluate MMAO-Dyn on an 18-scenario synthetic dynamic continuous benchmark matrix covering shifted sphere, shifted Ackley, and shifted Rastrigin landscapes at $10D$, $20D$, and $30D$, with two change severities and 12 seeds per scenario. The comparison layer includes a generic MMAO variant without dynamic derivation, dynamic random search, dynamic PSO-lite, dynamic DE-lite, and three endogenous ablations. Across the full 216-run matrix, MMAO-Dyn attains mean offline error $28.07$, improving over Generic-MMAO ($29.36$), Dynamic-PSO-lite ($34.65$), Dynamic-DE-lite ($67.09$), and Dynamic-RandomSearch ($111.37$). The gains are clearest in aggregate robustness on sphere and Rastrigin families and in 10-step post-change recovery relative to the generic backbone, whereas the seed-aligned comparison with Dynamic-PSO-lite remains unfavorable in win-loss count and the \texttt{NoMemoryRefresh} ablation stays very close to the full method. We therefore position MMAO-Dyn as a credible family-expansion result for MMAO: the metabolic loop can generate meaningful dynamic behavior, but the strongest current value lies in recovery-oriented resource redistribution rather than in universal dominance or in a fully optimized submechanism design.