Closed ZLI-afk closed 2 months ago
When I try to run energy minimization of HCP Ti by LAMMPS via DPA-1_OC2M, I found different prediction result with two different mapping scheme. I wonder which one is correct. (the alphabetic FULL mapping list is adopted according to input.json from https://www.aissquare.com/datasets/detail?pageType=datasets&name=Open_Catalyst_2020(OC20_Dataset))
LAMMPS
input.json
FULL list mapping: relax_task_full_list.tar.gz
Single element mapping: relax_task_single.tar.gz
v2.1.5
commit: 6e3d4a62
"image_name": "registry.dp.tech/dptech/prod-11045/phonolammps-dpmd:2.1.5"
As described in the summary section
Assigning the type map via pair_coeff (#2267) was added in v2.2.0.
pair_coeff
Summary
When I try to run energy minimization of HCP Ti by
LAMMPS
via DPA-1_OC2M, I found different prediction result with two different mapping scheme. I wonder which one is correct. (the alphabetic FULL mapping list is adopted according toinput.json
from https://www.aissquare.com/datasets/detail?pageType=datasets&name=Open_Catalyst_2020(OC20_Dataset))FULL list mapping: relax_task_full_list.tar.gz
Single element mapping: relax_task_single.tar.gz
DeePMD-kit Version
v2.1.5
Backend and its version
commit: 6e3d4a62
Python Version, CUDA Version, GCC Version, LAMMPS Version, etc
"image_name": "registry.dp.tech/dptech/prod-11045/phonolammps-dpmd:2.1.5"
Details
As described in the summary section