Data and inputs for the paper on surface diffusion on Ag (111). We highly recommend the data is downloaded as a zip file for use.
The trained DeepMD-kit model (graph.pb) can be used with Atomic Simulation Environment (ASE) as follows:
from deepmd.calculator import DP
DPcalculator=DP(model="graph.pb")
atoms.set_calculator(DPcalculator)Installation instructions for DeepMD-kit and ASE can be found on their websites.
Folder structure for data:
📦 data_for_paper
│
└───📂 data
│ │ Model Training, Testing and validation datasets separated by each adsorbate
│ ┣ 📜 type_map.json
│ │ ...
│ └───📂subfolders
│ │ ┣📂 test
│ │ | ┣ 📜box.npy
│ │ | ┣ 📜coord.npy
│ │ | ┣ 📜energy.npy
│ │ | ┣ 📜force.npy
│ │ ┣📂 train
│ │ | ┣ 📜box.npy
│ │ | ┣ 📜coord.npy
│ │ | ┣ 📜energy.npy
│ │ | ┣ 📜force.npy
│ │ ┣📂 validation
│ │ | ┣ 📜box.npy
│ │ | ┣ 📜coord.npy
│ │ | ┣ 📜energy.npy
│ │ | ┣ 📜force.npy
│
└───📂 lammps_example_in
│ ┣ 📜data.atoms (Example input for methyl on Ag(111))
│ ┣ 📜in.atoms (LAMMPS input file)
│
└───📂 train
│ │ Ensemble of trained models (4 models) differing only in their initial weights seeds
│ │
│ └───📂 00
│ │ ┣ 📜graph.pb ( Trained DeepMD model)
│ │ ┣ 📜in.json (Input parameters for training)
│ │ ┣ 📜lcurve.out (Loss curve output)
│ | ...
│
└───📂 vasp_example_in
│ ┣ 📜INCAR_ads (VASP input for AIMD on just adsorbates)
│ ┣ 📜INCAR_Ag_ads (VASP input for AIMD on adsorbates on Ag (111))