Reproduce “ADAPT for Science: A Multi-Agent System for Self-Evolving Scientific Optimization” 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
Zhaoran Feng, Yuchen Ni, Jianan Liu, Lemao Liu, Dejing Dou, Kairui Feng
ADAPT optimizes gravitational-wave detector designs by selecting historical starting points and adapting an Adam-based search using simulator feedback. It screens and probes starts within the same four-hour budget used for optimization, then applies a deadline-aware optimizer. Across eight held-out topologies, its feasible loss was 0.204 ± 0.101, lower than NAAdam and GEPA; evidence is limited to this differentiable simulator.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.