EDBench: Large-Scale Electron Density Data for Molecular Modeling

hongxin xiang, Ke Li, Mingquan Liu, Zhixiang Cheng, Bin Yao, Wenjie Du, Jun Xia, Li Zeng, Xin Jin, xiangxiang Zeng

Track: Track 2: Dataset Proposal Competition

Keywords: Molecular Geometry Learning, Electronic Density Database, Electronic Density Benchmark, Density Functional Theory

Abstract: We propose EDBench, a large-scale, high-quality dataset of electron density (ED) designed to advance learning-based research at the electronic scale. EDBench comprises 3,359,472 drug-like molecules with corresponding ED distributions and a comprehensive set of quantum chemical properties, including energy components, orbital energies, and multipole moments, thus providing a solid foundation for systematically investigating the role of ED in molecular modeling. We outline the AI tasks, data rationale, acceleration potential, and concrete pathway to data-creation with cost and scalability of EDBench.

Submission Number: 260