My Community

Dear all,

I have a question about the energy of system calculated by ML_FF.
Does it make sense, or be able to characterize the system, such as the surface energy, binding energy, etc?

Best

Shujun

Der Shujun,

the machine-learned force field is trained on forces and the total energy, so yes, differences in the total energy of the material can be used to characterize it. However, be sure that you have constructed your force field with your intended problem in mind. I.e., if you have no surface in your training data, you will not get a reasonable surface energy out of your force field.

If you have not already, please make yourself familiar with our ML_FF documentation on the wiki and also study the tutorial on machine learning force fields.

Don't hesitate to ask more specific questions if you have any. If not, please indicate that this answers your question so the topic can eventually be locked.

Cheers, Michael