[LAMMPS] DPLR potential can't give the outputs after the initial step was conducted. HELP!!!

[gzp1314]

the DPLR potential stops giving outputs after the initial step was done without any error meassages, altough i continue this jod for 5 hours. the training preformances of deep_wannier and DPLR model always give me confidence but the MD run was stuck after the initial step without any error messages giving off. this situation drives me crazy. the follwing is the tail of the output in lammps's logfile.
''
master list distance cutoff = 9.5
ghost atom cutoff = 9.5
binsize = 4.75, bins = 10 10 5
1 neighbor lists, perpetual/occasional/extra = 1 0 0
(1) pair deepmd, perpetual
attributes: full, newton on
pair build: full/bin
stencil: full/bin/3d
bin: standard
Setting up Verlet run ...
Unit style : metal
Current step : 0
Time step : 0.0005
Per MPI rank memory allocation (min/avg/max) = 11.35 | 11.35 | 11.35 Mbytes
Step Volume Temp PotEng KinEng TotEng Press
WARNING: Too many warnings: 291 vs 100. All future warnings will be suppressed (src/thermo.cpp:465)
0 39851.779 0 -1160.9988 0 -1160.9988 -8006.5184
''

PS : the lammps i used is the built-in version in the deepmd-kit (version = 2.1.5).
PS : I test my environment by sucessfully runing the example of water given by the hands-on sessions on DPLR (this is the link of it : [https://github.com/CSIprinceton/workshop-july-2022/tree/main/hands-on-sessions])

[wanghan-iapcm]

Could you please provide your lammps input script?

[gzp1314]

Thanks for your reply.
You can find the configuration file and lammps script in the attachment. I also put the training scripts about the deep_wannier and DPLR model in it.

Update:
I just fail to get the result when i use the follwing python script to check the dipole distribution of my atomic configuration by my deep_wannier model. This information may be help.
"
from deepmd.infer import DeepDipole
import ase.io.vasp
import numpy as np

f = './3/4168/'
dipole_model = DeepDipole('./diople.pb')

configs = ase.io.vasp.read_vasp(f+"/POSCAR")

atypes = np.array(configs.get_atomic_numbers()) - 1

dipole = dipole_model.eval(configs.get_positions(),configs.get_cell(),atom_types=atypes)

"
SCRIPTS.zip

[wanghan-iapcm]

By your script you want to associate virtual atom type 3 to real atom type 2, but the .lmp file is linking atoms type3 to type 1.

PS: We recommend setting the coordinates of the virtual atoms the same as the real atoms to which it associated.

[gzp1314]

I must say sorry for my mistake to provide a wrong version of data.lmp. I already corrcted this few days ago. However, the same question still. This is the current data.lmp. Unfortunately, with this data file and the same lammps script, the MD still refuse to give me output after the initial step.

PS : The purple stick is represent for the bonds. Now the virtual atom (type 3) correctly associate to the real atom (type 2 )

"

LAMMPS data file written by OVITO Basic 3.7.4

54 atoms
3 atom types
18 bonds
1 bond types
0.0 14.323764456 xlo xhi
0.0 15.456744154 ylo yhi
0.0 20.0 zlo zhi
0.0 0.015286353 0.006080383 xy xz yz

Masses

1 207.2 # Pb
2 127.6 # Te
3 127.6 # X

Atoms # full

1 1 1 2 13.94989 2.15493 8.51626
2 1 1 2 0.24207 7.26525 8.58236
3 1 1 2 0.15916 13.4256 9.20057
4 1 1 2 5.54125 2.98099 8.54042
5 1 1 2 4.26232 7.2625 8.37701
6 1 1 2 4.87457 12.91533 8.7699
7 1 1 2 9.58226 2.56973 8.4505
8 1 1 2 9.74629 7.89716 8.78916
9 1 1 2 9.42717 11.88904 8.47521
10 1 1 2 3.02151 0.46799 11.32202
11 1 1 2 2.09971 4.462 11.43385
12 1 1 2 2.56724 10.01677 11.69191
13 1 1 2 7.60635 15.3216 11.26812
14 1 1 2 7.35365 5.54337 10.98555
15 1 1 2 6.63247 9.60885 11.70394
16 1 1 2 11.56079 14.80505 11.57819
17 1 1 2 12.16843 5.12687 11.53209
18 1 1 2 12.30139 10.35047 11.40212
19 1 2 6 14.03871 2.47346 11.55175
20 1 2 6 0.11287 8.11184 11.57619
21 1 2 6 13.95946 13.00409 12.16964
22 1 2 6 5.00136 3.22749 11.55536
23 1 2 6 4.36581 7.41253 11.39038
24 1 2 6 5.00799 13.50131 11.71275
25 1 2 6 9.55782 3.23548 11.3682
26 1 2 6 9.20674 8.0407 11.76647
27 1 2 6 9.21337 12.705 11.35689
28 1 2 6 2.31962 0.49252 8.37993
29 1 2 6 2.24438 4.95096 8.41343
30 1 2 6 2.33173 11.01319 8.82926
31 1 2 6 7.28501 0.43129 8.23732
32 1 2 6 6.97837 5.80314 7.94196
33 1 2 6 6.80557 10.31466 8.74035
34 1 2 6 11.78998 15.36115 8.51581
35 1 2 6 11.94364 5.56808 8.55297
36 1 2 6 12.0343 10.37308 8.41946
37 1 3 -8 14.08434134 2.64653618 11.36588609
38 1 3 -8 0.128973876 8.213134505 11.40918083
39 1 3 -8 13.81206715 13.01521547 11.905815
40 1 3 -8 4.886063683 3.19141205 11.34439811
41 1 3 -8 4.407133588 7.57302667 11.23523174
42 1 3 -8 5.013297788 13.65286493 11.48333655
43 1 3 -8 9.565371448 3.338126793 11.15672568
44 1 3 -8 9.067783288 8.020554828 11.54011384
45 1 3 -8 9.2735376 12.86216585 11.19784527
46 1 3 -8 2.206231955 0.468200139 8.6252808
47 1 3 -8 2.233673708 5.125573938 8.59355138
48 1 3 -8 2.390679628 11.10217932 9.024382908
49 1 3 -8 7.319796113 0.61674052 8.449932988
50 1 3 -8 6.875238583 5.742005995 8.183818348
51 1 3 -8 6.86667362 10.4088793 8.90348149
52 1 3 -8 11.86036383 15.47757568 8.703367773
53 1 3 -8 11.9924759 5.682586195 8.75019083
54 1 3 -8 11.85851064 10.36812398 8.63789749

Bonds

1 1 19 37
2 1 20 38
3 1 21 39
4 1 22 40
5 1 23 41
6 1 24 42
7 1 25 43
8 1 26 44
9 1 27 45
10 1 28 46
11 1 29 47
12 1 30 48
13 1 31 49
14 1 32 50
15 1 33 51
16 1 34 52
17 1 35 53
18 1 36 54

"

[wanghan-iapcm]

Please try setting the coordinates of the virtual atoms the same as the real atoms to which it associated.

[gzp1314]

Thank you very much !!!. I'm so happy it is fixd after i set the same coordinates about the real atoms and the corrsponding virtual atoms.
But i'm still wondering why it must be the same coordinates between the real atoms and the corrsponding virtual atoms. This constraint feels a little SPECIAL

[wanghan-iapcm]

The virutal atoms should be located to the same subdomain as its associated real atom during the domain decomposition. This can only be safely guaranteed if they have the same coordinate.

[gzp1314]

Thanks, it helps me a lot.