Confuse with the dpgen init_surf procedure

[ysu-007]

Summary

DPGEN Version and Platform

Job submission and computing cluster configuration

Details

In the dpgen manual, there are only two steps for the dpgen init_surf:
_**Basically init_surf can be devided into two parts , denoted as stages in PARAM:

1. Build specific surface in folder 00.place_ele
2. Pertub and scale in folder 01.scale_pert_**_

For the first step, the software generates some initial surfaces with a vacuum and calculates their energy with DFT method. Then the second step will give perturbation and scale to the initial structure generated in the first step, so are there any DFT calculations for the new structures generated in the second step? I don't find any DFT calculation yet in my test. For the dpgen init_bulk, I know there will be AIMD calculations for the structures generated in the second step which is step 3. However, there is no step 3 in init_surf.

[ysu-007]

What I see from the code is that the init_surf generates a lot of structures but doesn't calculate the DFT energy. Does the dpgen do the DFT calculation in another process?

[lnnbig]

Dear Huang @HuangJiameng , I really appreciate you for your helpful explanation. For my understanding from your words, I just need to prepare the fparam.raw, which consists of the temperature of each frame, and set the temperature values in Exploration step. The SIGMA setting in INCAR for 02.fp is doesn't matter, right? btw, the unit of temperature values in fparam.raw should be K or eV? Please excuse my rambling

I have also raised my question in DeePMD. Please disregard the message below if you come across it there.

May I inquire if you have resolved the fparam issue? I am also confused by it.

Firstly, according to the file found at this link: https://github.com/deepmodeling/deepmd-kit/blob/master/examples/fparam/lmp/in.lammps, the line "pair_style deepmd frozen_model.pb fparam 0.25852028" suggests that the unit for fparam is clearly expressed in electron volts (eV). However, when I generated the Lammps file using DP-GEN, the unit indicated was in Kelvin (K).

Furthermore, there seems to be a significant problem with the value of max_devi_f in the model_devi.out file, which is remarkably small (1e-14). This suggests that there may be bugs present in both DPGEN (Version: 0.9.2) and DeePMD (v2.0.1).

[phyoung123]

No 1. The electronic temperature eV in lammps input file do not conflict with fparam, in which, However, actually is K, you can check it in DPGEN auto-generated file after Labeling step.
No 2. I don't really know, maybe you can update your DPGEN version.

[lnnbig]

No 1. The electronic temperature eV in lammps input file do not conflict with fparam, in which, However, actually is K, you can check it in DPGEN auto-generated file after Labeling step. No 2. I don't really know, maybe you can update your DPGEN version.

I double checked the fparam,npy in the example at this link: https://github.com/deepmodeling/deepmd-kit/tree/master/examples/fparam/data/e3000_i2000/set.000, which is showed below:

np.load('fparam.npy')
array([0.25852028, 0.25852028, 0.25852028, 0.25852028, 0.25852028,
0.25852028, 0.25852028, 0.25852028, 0.25852028, 0.25852028,
0.25852028, 0.25852028, 0.25852028, 0.25852028, 0.25852028,
0.25852028, 0.25852028, 0.25852028, 0.25852028, 0.25852028,
0.25852028, 0.25852028, 0.25852028, 0.25852028, 0.25852028,
0.25852028, 0.25852028, 0.25852028, 0.25852028, 0.25852028,
0.25852028, 0.25852028, 0.25852028, 0.25852028, 0.25852028,
0.25852028, 0.25852028, 0.25852028, 0.25852028, 0.25852028,
0.25852028, 0.25852028, 0.25852028, 0.25852028, 0.25852028,
0.25852028, 0.25852028, 0.25852028, 0.25852028, 0.25852028,
0.25852028, 0.25852028, 0.25852028, 0.25852028, 0.25852028,
0.25852028, 0.25852028, 0.25852028, 0.25852028, 0.25852028,
0.25852028, 0.25852028, 0.25852028, 0.25852028, 0.25852028,
0.25852028, 0.25852028, 0.25852028, 0.25852028, 0.25852028,
0.25852028, 0.25852028, 0.25852028, 0.25852028, 0.25852028,
0.25852028, 0.25852028, 0.25852028, 0.25852028, 0.25852028,
0.25852028, 0.25852028, 0.25852028, 0.25852028, 0.25852028,
0.25852028, 0.25852028, 0.25852028, 0.25852028, 0.25852028,
0.25852028, 0.25852028, 0.25852028, 0.25852028, 0.25852028,
0.25852028, 0.25852028, 0.25852028, 0.25852028, 0.25852028],
dtype=float32)

which is in eV, not in K.

[lnnbig]

No 1. The electronic temperature eV in lammps input file do not conflict with fparam, in which, However, actually is K, you can check it in DPGEN auto-generated file after Labeling step. No 2. I don't really know, maybe you can update your DPGEN version.

Thank you very much for your reply. May I please ask which version of DP-GEN are you using?

[AnguseZhang]

There're several definitions to be clarifed.

1. There's no need to care about the unit offparam, as long as the unit is consistent in all data you trained. So its unit can either be eV or K, or other units according to its physical meaning. Since it only indicates an additional input parameter of neuron network, apart of coordinates.
2. The energy unit in deep potential is eV, and force unit is eV/A.
3. In lammps, the temperature unit is K, so what you see in LAMMPS may be the unit of temperature.