Reproduce “Online Automated Algorithm Design with Large Language Models” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Zhiyao Zhang, Yichen Li, Xingyu Wu, Liang Feng, Kay Chen Tan
OnDesign redesigns executable search logic from optimization-state feedback as a black-box run unfolds, using state analysis, multi-agent deliberation, and periodic guideline refinement for Bayesian and evolutionary optimization. Across 17 tested suite–dimension configurations, it has the best average rank; on 30-dimensional BBOB, its rank is 2.02 versus 5.10 for FunBO. Evidence covers single-objective problems up to 30 dimensions.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.