Reproduce “ASPIRE: An Agentic System-Identification Platform for Iterative Reasoning and Exploration” 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
Erik Derner, Jiří Kubalík, Robert Babuška
ASPIRE identifies closed-form models from measured data using a language-model agent, while a separate harness evaluates predictions and checks physical requirements. On a dry-friction system, matched PySR had a lower mean test RMSE than ASPIRE, with no observed difference among three approaches. All 12 comparison-tool models violated the equilibrium requirement; ASPIRE candidates had no demonstrated violations, though most checks were inconclusive.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.