Title: NIST’s JARVIS-ML Repository Accelerates Materials Discovery
NIST has developed JARVIS-ML, a repository of machine learning models and data designed to accelerate materials discovery. The repository includes Classical Force-field Inspired Descriptors (CFID), a universal framework for representing materials’ chemistry, structure, and charge-related data. JARVIS-ML has developed high-accuracy ML models for various material properties, including formation energies, bandgaps, and magnetic moments.
The repository also includes STM-image ML models, which can directly accelerate experiments. Graph convolution neural network models are being developed for image and crystal structure analysis. By providing transparent workflows, datasets, and models, JARVIS-ML aims to enhance the reproducibility and efficiency of materials discovery research.
Source: https://www.nist.gov/programs-projects/jarvis-ml
Keywords: Machine, Descriptors, Framework, Quantum, Models
