MedeA 3.4 includes groundbreaking Machine-Learned Potential (MLP) capabilities, including neural network-based potential models and a fully integrated MLP generator. With these capabilities, users can create highly accurate MLP descriptions of ab initio training sets for use with the LAMMPS simulation environment. These enhancements allow users to simulate systems with first-principles accuracy with substantial system sizes and simulation timescales which greatly exceed those accessible to quantum mechanical methods.
'With every release we increase scientific capabilities, ease-of-use, and efficiency for MedeA for materials modelers.' said Dr. Benoit Leblanc, Chief Software Architect at Materials Design. 'With the MedeA 3.4 release, in addition to providing a unique and efficient Machine-Learned Potential Generator capability, we have significantly updated many of the views and controls that comprise the MedeA interface. These updates make MedeA even more capable, and its interface even more intuitive, responsive, and efficient. From a software design perspective, we are very proud of these enhancements, they provide value to users immediately and, for the programming team, such enhancements facilitate the future development of MedeA.'
MedeA 3.4 features substantial user interface updates, which enhance productivity and ease-of-use. Additionally, many graphical updates have been made, including enhanced display modes and graphical analysis capabilities. Updates have been made to the VASP interface to facilitate the creation and analysis of large training sets. A bulleted list of major enhancements is provided below.
Dr. Leblanc added, 'We are committed to serving the expanding community of materials simulators who use the MedeA environment. I am truly delighted to see the output of our software development work having an important effect on materials research in many organizations worldwide and I look forward to hearing your comments and suggestions for future releases.'