Closed duousername1 closed 3 weeks ago
It seems that this error message has been removed in https://github.com/deepmodeling/abacus-develop/pull/4548. ping @pxlxingliang for confirmation.
hi @duousername1, ntype
is unnecessary now, please remove this parameter in your setting.
It seems that this error message has been removed in deepmodeling/abacus-develop#4548. ping @pxlxingliang for confirmation. Thank you, I will try following the method you provided.
hi @duousername1,
ntype
is unnecessary now, please remove this parameter in your setting. Thank you, I will try following the method you provided.
hi @duousername1,
ntype
is unnecessary now, please remove this parameter in your setting.
hi
The suggestion provided above helps address the issue with the number of atomic species, but I have encountered another problem. While performing single-point energy calculations during the iterations, I received the following error: "NLOCAL < NBANDS. CHECK IN FILE: OUT.ABACUS/warning.log." Therefore, I would like to set different parameters (such as the value of nbands
) separately for different structures. Could you please advise on how to configure this in the param.json
file? thanks
@njzjz @pxlxingliang
Hi @duousername1, Please do not set nbands
in param.json
, which is different for different systems. The error is because the parallel cores are larger than band number, you should use a small machine with less parallel cores.
NLOCAL < NBANDS. CHECK IN FILE: OUT.ABACUS/warning.log.
Sorry, I have made a mistake, the NLOCAL is the basis number. Just removing the setting of nbands
in param.json
will be fine.
NLOCAL < NBANDS
Thank you again, it worked successfully! @pxlxingliang
Summary
Hello, I have a question. During iteration for single-point energy calculations, some calculations are successful, while others fail. The reason for the failures is that the number of atom types in the INPUT does not match the number of atom types in the structure file. This might be because I set
ntype=6
in theparam.json
file, but in the dataset for single-point energy calculation, there is a structure like NH4ClO4 wherentype=4
, so they don't match. How do people usually handle this kind of situation? The reason I setntype=6
is that some of my structures consist of 6 atom types.DP-GEN Version
0.12.0
Platform, Python Version, etc
"api_version": "1.0", "deepmd_version": "2.2.8", bohrium paltform
Details