Can an LLM learn how an optimizer behaves -- and use that knowledge to control it? We extend Code World Models (CWMs), LLM-synthesized Python programs that predict environment dynamics, from deterministic games to stochastic combinatorial optimization. Given suboptimal trajectories of (1+1)-RLSk, the LLM synthesizes a simulator of the optimizer's dynamics; greedy planning over this simulator then selects the mutation strength k at each step. On \lo{} and \onemax{}, CWM-greedy performs within 6\% of the theoretically optimal policy -- without ever seeing optimal-policy trajectories. On \jump{k}, where a deceptive valley causes all adaptive baselines to fail (0\% success rate), CWM-greedy achieves 100\% success rate -- without any collection policy using oracle knowledge of the gap parameter. On the NK-Landscape, where no closed-form model exists, CWM-greedy outperforms all baselines across fifteen independently generated instances (36.94 vs.\ 36.32; p<0.001) when the prompt includes empirical transition statistics. The CWM also outperforms DQN in sample efficiency (200 offline trajectories vs.\ 500 online episodes), success rate (100\% vs.\ 58\%), and generalization (k=3: 78\% vs.\ 0\%). Robustness experiments confirm stable synthesis across 5 independent runs.
@article{arxiv.2602.22260,
title = {Code World Models for Parameter Control in Evolutionary Algorithms},
author = {Camilo Chacón Sartori and Guillem Rodríguez Corominas},
journal= {arXiv preprint arXiv:2602.22260},
year = {2026}
}