Skip to content

Very bad performance for molecular crystals. #15

@husakm

Description

@husakm

I had run AIM2Net on crystal structure of 50 molecular crystals. I had than compared the RMSCD with experiment and other methods. The results were very bad - worse than p-GNFF, MACE and as expected DFT. Maybe the defalut AIM2Net setting is not suitable for molecular crystals ? Any advice for setup is welcome.
Should I report to autors the worse outliers for analysis ?

Related question - you already implemented periodic Ewald electostatic sumation (as mentioned in 7-th blind test supplementary materials). Any change to make it possible to activate ?

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions