GAP-ML model, training and MD dataset for "A Composition-Transferable Machine Learning Potential for LiCl-KCl Molten Salts Validated by HEXRD"

Guo, Jicheng; Ward, Logan; Babuji, Yadu; Hoyt, Nathaniel; Williamson, Mark; Foster, Ian; Jackson, Nicholas; Benmore, Chris; Sivaraman, Ganesh


MDF Open



Source Name



CC-BY 4.0



10.18126/66pj-gpnr View on Datacite
Usage Notes ML training dataset and potential 1) "POT/" MD
  1. The full MD trajectory for the thermal conducitivity calculations can be found in "MD_WAVE_METHOD/"
  2. All the MD trajectories used for structure factor / CN estimation at multiple composition / temperatures can be foudn in "MD/". The ".tar.gz" file with '.extxyz' trajectory is according to the LiCl-KCl molar fraction followed by the temperature in K units.
  3. The "MD/" folder also the log files for two additional compositions (i.e. 30-70, & 80-20) and can be used to verify the density from volume relaxation.
Misc. Info Molten LiCl as a system on its own is reported in The potential provided in this data deposit does not include any pure LiCl melt training data. Hence should be used with caution at ultra low concentrations of LiCl.