Reproduce “Risk-Controlled DFT Selection for Machine-Learned Interatomic Potentials” 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
Kun Ryu, Yeo Jin Jung, Shyue Ping Ong, Claire Donnat, Ying Shirley Meng
The authors develop a structure-based score to identify molecular configurations whose machine-learned interatomic potential (MLIP) predictions may need replacement by density functional theory (DFT). A conformal risk-control (CRC) procedure calibrates the routing threshold using labeled calibration data. At target α = 0.30, the method routes 57.8% of configurations to DFT, fewer than two committee baselines, under an expectation-level capped-loss guarantee.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.