Reproduce “Learning to Outgrow a Theory: Experimental Discovery Beyond the Initial Hypothesis Space” 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

SiYuan Ma, Albert Gao, Chunzheng Zhu, Xin Yan, Wenlong Zhang, Wenxin Zhang, Luqi Gong, Tianlin Li and 1 more

The authors introduce a system pairing structural edits with experiments to test them. It revises a class only when sequential evidence rejects it, and can flag when its edit library lacks a supported edit. Across 400 environments, it recovered mechanisms in 89.5% at a budget of 32, versus 79.5% for independent selection; recovery is limited to mechanisms expressible by its edit grammar.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.