Hi julian,
I’m sorry to bother you. Currently, I am learning how to fit potential. In official examples, it is generally required to have certain initial values. If there is no reference in the literature for these initial values, how can I obtain them? Additionally, when setting the observed values, how should different energies be specified, such as cohesive energy, surface energy, attachment energy, and so on? Thank you for your assistance.
Sincerely,
LiMa
Hi LiMa,
If there are no literature values you just have to guess some initial ones. Just look for similar systems (particularly similar charges) and use those as starting points. For energies, GULP allows you to fit the energy different between configurations, or the absolute value of a single configuration. So you just need to include configurations so that you can create the reaction of interest and fit the difference. This will handle cohesive and surface energies (though normally you wouldn’t fit these), but not attachment. At the moment there isn’t option to fit these & no one has ever wanted to (until now).
Regards,
Julian
Hi julian,
Thank you very much for your reply. I apologize for bothering you again. I have reviewed the explanation of the reaction command in the help document, but I am not very clear about the meaning represented by the specific parameters listed in it which as shown in the figure below, and if I need to fit other relative energies, such as surface energy, do I need to calculate the initial and final state energies of the system based on the surface energy calculation formula, and then substitute them into the reaction command, And then fit the surface energy? In addition, are there any specific requirements for setting the weight command value?
Thank you for your assistance.
Sincerely,
LiMa
By definition, a reaction energy is the difference in energy between the initial and final states & so indeed these are what they should be. The weight is something you can choose based on the magnitude of the energy difference & how important it is to get it right (relative to other observables).