-
Hi, i recently want to do some test about the accuracy of CHGNet, so i choose some structures and test the difference of energies predicted by CHGNet and VASP. The result is shown below. You can find it's not very close. CHGNet always obtains a lower energy. I don't know why, maybe i used different settings in VASP?
Below is the CHGNet code and INCAR i use. Please give me some advice. Thanks so much!
|
Beta Was this translation helpful? Give feedback.
Answered by
bowen-bd
Dec 5, 2024
Replies: 2 comments 2 replies
-
|
|
Beta Was this translation helpful? Give feedback.
2 replies
-
|
This is due to the MP comparability. CHGNet energies are MPRelaxSet+compatibility energies. https://matsci.org/t/mp-compatibility-and-energy-corrections/35035 |
Beta Was this translation helpful? Give feedback.
Answer selected by
bowen-bd
-
|
Thank you so much! |
Beta Was this translation helpful? Give feedback.
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
This is due to the MP comparability. CHGNet energies are MPRelaxSet+compatibility energies.
Check
https://matsci.org/t/mp-compatibility-and-energy-corrections/35035