MMAO Optimizer Tops PSO-lite and ES-lite in Fair-Budget Benchmarks

A new empirical evaluation from Jinliang Xu and Liping Ma puts the Metabolic Multi-Agent Optimizer (MMAO) through its paces under a strict fair-budget protocol. The study moves beyond re-proposing the framework to ask whether MMAO's closed-loop resource-allocation principle holds water across broader, standard benchmarks. On the continuous side, the team tested 8 CEC2017 functions at 10D and 30D with 20 random seeds each. For discrete optimization, they ran 5 TSPLIB traveling-salesman instances (also 20 seeds) plus an auxiliary OR-Library multiple-knapsack slice. Baselines included PSO-lite, ES-lite, and an iterated-greedy 2-opt route heuristic, all with matched computational budgets. Trajectory-level diagnostics tracked communal budget usage, success rates, role evolution, and population turnover.

The results show MMAO clearly beating the external baselines on both continuous and TSPLIB benchmarks. Ablation variants of MMAO remained much closer to the full method than any external baseline, confirming that the core mechanism—endogenous resource redistribution under evidence pressure—drives performance. The authors note that the biggest remaining gap isn't basic workability but sharper mechanism isolation and broader competition-grade comparison. Still, the paper positions MMAO as a benchmark-backed cross-domain adaptive framework, with validated value in dynamic resource allocation for complex optimization tasks. For practitioners, this signals a strong alternative to traditional swarm or evolution strategies when budget fairness is critical.