Hello everyone,
I am a beginner using dpgen. I trained a model for fcc-Al. However, when I tested the model, I got a significant deviation for self-interstitial formation energy. Could anyone please provide me a few suggestions? The input files and results are listed below,
param.json in init_bulk
{
"stages": [1, 2, 3, 4],
"cell_type": "fcc",
"super_cell": [2, 2, 2],
"elements": ["Al"],
"from_poscar": true,
"from_poscar_path": "POSCAR",
"potcars": ["POTCAR"],
"relax_incar": "INCAR.rlx",
"md_incar": "INCAR.md",
"scale": [1.00],
"skip_relax": false,
"pert_numb": 50,
"md_nstep": 20,
"pert_box": 0.03,
"pert_atom": 0.01,
"coll_ndata": 5000,
"type_map": ["Al"],
"_comment": "that's all"
}
param.json in run
{
"type_map": ["Al"],
"mass_map": [27],
"init_data_prefix": "../init/",
"init_data_sys": [
"POSCAR.02x02x02/02.md/sys-0032/deepmd"
],
"init_batch_size": [
1
],
"sys_configs": [
["/home/zhq/WORK/fzh/work/ml/test/Al-2/init/POSCAR.02x02x02/01.scale_pert/sys-0032/scale-1.000/00000[0-4]/POSCAR"],
["/home/zhq/WORK/fzh/work/ml/test/Al-2/init/POSCAR.02x02x02/01.scale_pert/sys-0032/scale-1.000/00000[5-9]/POSCAR"],
["/home/zhq/WORK/fzh/work/ml/test/Al-2/init/POSCAR.02x02x02/01.scale_pert/sys-0032/scale-1.000/00001*/POSCAR"],
["/home/zhq/WORK/fzh/work/ml/test/Al-2/init/POSCAR.02x02x02/01.scale_pert/sys-0032/scale-1.000/00002*/POSCAR"],
["/home/zhq/WORK/fzh/work/ml/test/Al-2/init/POSCAR.02x02x02/01.scale_pert/sys-0032/scale-1.000/00003*/POSCAR"],
["/home/zhq/WORK/fzh/work/ml/test/Al-2/init/POSCAR.02x02x02/01.scale_pert/sys-0032/scale-1.000/00004*/POSCAR"]
],
"_comment": " 00.train ",
"numb_models": 4,
"default_training_param" : {
"model":{
"_comment": " model parameters",
"type_map":["Al"],
"descriptor":{
"type": "se_a",
"sel": [200],
"rcut_smth": 0.5,
"rcut": 6.0,
"neuron": [25, 50, 100],
"resnet_dt": false,
"axis_neuron": 12,
"seed": 1
},
"fitting_net":{
"neuron": [240, 240, 240],
"resnet_dt": true,
"sedd": 1
}},
"learning_rate":{
"type": "exp",
"start_lr": 0.001,
"decay_steps": 2000,
"decay_rate": 0.95
},
"loss":{
"start_pref_e": 0.02,
"limit_pref_e": 2,
"start_pref_f": 1000,
"limit_pref_f": 1,
"start_pref_v": 0.0,
"limit_pref_v": 0.0
},
"training":{
"coord_norm": true,
"type_fitting_net": false,
"_comment": " traing controls",
"systems": [],
"set_prefix": "set",
"stop_batch": 400000,
"batch_size": 1,
"seed": 0,
"_comment": " display and restart",
"_comment": " frequencies counted in batch",
"disp_file": "lcurve.out",
"disp_freq": 2000,
"numb_test": 4,
"save_freq": 20000,
"save_ckpt": "model.ckpt",
"load_ckpt": "model.ckpt",
"disp_training": true,
"time_training": true,
"profiling": false,
"profiling_file": "timeline.json",
"_comment": "that's all"}
},
"_comment": " 01.model_devi ",
"_comment": "model_devi_skip: the first x of the recorded frames",
"model_devi_dt": 0.002,
"model_devi_skip": 0,
"model_devi_f_trust_lo": 0.05,
"model_devi_f_trust_hi": 0.20,
"model_devi_e_trust_lo": 1e10,
"model_devi_e_trust_hi": 1e10,
"model_devi_clean_traj": true,
"model_devi_jobs":
[
{
"_idx": 0,
"ensemble": "npt",
"nsteps": 300,
"press": [
1.0, 10, 100
],
"sys_idx": [
0
],
"temps": [
50
],
"trj_freq": 10
},
{
"_idx": 1,
"ensemble": "npt",
"nsteps": 1000,
"press": [
1.0, 10, 100
],
"sys_idx": [
0, 1
],
"temps": [
50
],
"trj_freq": 10
},
{
"_idx": 2,
"ensemble": "npt",
"nsteps": 1000,
"press": [
1.0, 10, 100
],
"sys_idx": [
2, 3
],
"temps": [
50
],
"trj_freq": 10
},
{
"_idx": 3,
"ensemble": "npt",
"nsteps": 3000,
"press": [
1.0, 10, 100
],
"sys_idx": [
4, 5
],
"temps": [
50
],
"trj_freq": 10
},
{
"_idx": 4,
"ensemble": "npt",
"nsteps": 3000,
"press": [
1.0, 10, 100
],
"sys_idx": [
4, 5
],
"temps": [
50
],
"trj_freq": 10
}
],
"_comment": " 02.fp ",
"fp_style": "vasp",
"shuffle_poscar": false,
"fp_task_max": 300,
"fp_task_min": 5,
"fp_pp_path": "./",
"fp_pp_files": ["POTCAR"],
"fp_incar": "/home/zhq/WORK/fzh/work/ml/test/Al-2/run/INCAR",
"_comment": " that's all "
}
param.json in test
{
"_comment": "models",
"potcar_map": {
"Al": "potential/POTCAR"
},
"conf_dir": "confs/Al/std-fcc",
"key_id": "",
"task_type": "deepmd",
"task": "all",
"vasp_params": {
"ecut": 650,
"ediff": 1e-6,
"kspacing": 0.15,
"kgamma": false,
"npar": 1,
"kpar": 1,
"_comment": " that's all "
},
"lammps_params": {
"model_dir": "Al_model",
"type_map": [
"Al"
],
"model_name": false,
"model_param_type": false
},
"_comment": "00.equi",
"alloy_shift": false,
"_comment": "01.eos",
"vol_start": 12,
"vol_end": 22,
"vol_step": 0.5,
"_comment": "02.elastic",
"norm_deform": 2e-2,
"shear_deform": 5e-2,
"_comment": "03.vacancy",
"supercell": [
3,
3,
3
],
"_comment": "04.interstitial",
"insert_ele": [
"Al"
],
"reprod-opt": false,
"_comment": "05.surface",
"min_slab_size": 10,
"min_vacuum_size": 11,
"_comment": "pert xz to work around vasp bug...",
"pert_xz": 0.01,
"max_miller": 2,
"static-opt": false,
"relax_box": false,
"_comment": "06.phonon",
"supercell_matrix": [
2,
2,
2
],
"band": "0 1 0 0.5 1 0.5 0.375 0.75 0.375 0 0 0 0.5 0.5 0.5",
"_comment": "that's all"
}
the result log
DeepModeling
Version: 0.8.2.dev0+gf8d70a4.d20210204
Date: Feb-04-2021
Path: /home/zhq/.local/lib/python3.7/site-packages/dpgen
Dependency
numpy 1.20.3 /apps/lib/anaconda/anaconda3/e5/lib/python3.7/site-packages/numpy
dpdata 0.1.19 /apps/lib/anaconda/anaconda3/e5/lib/python3.7/site-packages/dpdata-0.1.19-py3.7.egg/dpdata
pymatgen 2020.10.9.01 /apps/lib/anaconda/anaconda3/e5/lib/python3.7/site-packages/pymatgen
monty 4.0.2 /apps/lib/anaconda/anaconda3/e5/lib/python3.7/site-packages/monty
ase 3.19.1 /apps/lib/anaconda/anaconda3/e5/lib/python3.7/site-packages/ase
paramiko 2.7.2 /home/zhq/.local/lib/python3.7/site-packages/paramiko
custodian 2021.1.8 /home/zhq/.local/lib/python3.7/site-packages/custodian
Reference
Please cite:
Yuzhi Zhang, Haidi Wang, Weijie Chen, Jinzhe Zeng, Linfeng Zhang, Han Wang, and Weinan E,
DP-GEN: A concurrent learning platform for the generation of reliable deep learning
based potential energy models, Computer Physics Communications, 2020, 107206.
Description
/gpfs01/zhq_work/fzh/work/ml/test/Al-2/test/00.equi/Al/std-fcc/deepmd
conf_dir: EpA(eV) VpA(A^3)
confs/Al/std-fcc -3.7483 16.503
generate /gpfs01/zhq_work/fzh/work/ml/test/Al-2/test/01.eos/Al/std-fcc/deepmd/vol-12.00
generate /gpfs01/zhq_work/fzh/work/ml/test/Al-2/test/01.eos/Al/std-fcc/deepmd/vol-12.50
generate /gpfs01/zhq_work/fzh/work/ml/test/Al-2/test/01.eos/Al/std-fcc/deepmd/vol-13.00
generate /gpfs01/zhq_work/fzh/work/ml/test/Al-2/test/01.eos/Al/std-fcc/deepmd/vol-13.50
generate /gpfs01/zhq_work/fzh/work/ml/test/Al-2/test/01.eos/Al/std-fcc/deepmd/vol-14.00
generate /gpfs01/zhq_work/fzh/work/ml/test/Al-2/test/01.eos/Al/std-fcc/deepmd/vol-14.50
generate /gpfs01/zhq_work/fzh/work/ml/test/Al-2/test/01.eos/Al/std-fcc/deepmd/vol-15.00
generate /gpfs01/zhq_work/fzh/work/ml/test/Al-2/test/01.eos/Al/std-fcc/deepmd/vol-15.50
generate /gpfs01/zhq_work/fzh/work/ml/test/Al-2/test/01.eos/Al/std-fcc/deepmd/vol-16.00
generate /gpfs01/zhq_work/fzh/work/ml/test/Al-2/test/01.eos/Al/std-fcc/deepmd/vol-16.50
generate /gpfs01/zhq_work/fzh/work/ml/test/Al-2/test/01.eos/Al/std-fcc/deepmd/vol-17.00
generate /gpfs01/zhq_work/fzh/work/ml/test/Al-2/test/01.eos/Al/std-fcc/deepmd/vol-17.50
generate /gpfs01/zhq_work/fzh/work/ml/test/Al-2/test/01.eos/Al/std-fcc/deepmd/vol-18.00
generate /gpfs01/zhq_work/fzh/work/ml/test/Al-2/test/01.eos/Al/std-fcc/deepmd/vol-18.50
generate /gpfs01/zhq_work/fzh/work/ml/test/Al-2/test/01.eos/Al/std-fcc/deepmd/vol-19.00
generate /gpfs01/zhq_work/fzh/work/ml/test/Al-2/test/01.eos/Al/std-fcc/deepmd/vol-19.50
generate /gpfs01/zhq_work/fzh/work/ml/test/Al-2/test/01.eos/Al/std-fcc/deepmd/vol-20.00
generate /gpfs01/zhq_work/fzh/work/ml/test/Al-2/test/01.eos/Al/std-fcc/deepmd/vol-20.50
generate /gpfs01/zhq_work/fzh/work/ml/test/Al-2/test/01.eos/Al/std-fcc/deepmd/vol-21.00
generate /gpfs01/zhq_work/fzh/work/ml/test/Al-2/test/01.eos/Al/std-fcc/deepmd/vol-21.50
Vpa(A^3) EpA(eV)
11.9999981608503 -2.96662795170267
12.4999949243073 -2.97802583445048
12.9999983614591 -3.09448430830428
13.4999983961358 -3.32230518208628
13.9999959870801 -3.5283274458636
14.5000040385406 -3.6320900440121
15.0000041246356 -3.6964484536534
15.5000045857403 -3.7304097714782
16.0 -3.74457075952198
16.4999997474047 -3.74831572706045
16.9999987265498 -3.74521051825265
17.4999961808157 -3.73651305151505
18.0000045936601 -3.7231315655221
18.4999940008706 -3.7045569241317
19.0000055945374 -3.67865231543475
19.5000043580999 -3.64584828796787
19.9999947077013 -3.6097276969094
20.4999931235139 -3.57066598363155
20.9999979876931 -3.52542444079503
21.5000005488178 -3.4737049576875
gen with norm [-0.02, -0.01, 0.01, 0.02]
gen with shear [-0.05, -0.025, 0.025, 0.05]
111.73 50.42 50.42 0.00 0.00 0.00
50.42 111.73 50.42 0.00 0.00 0.00
50.42 50.42 111.73 0.00 0.00 0.00
0.00 0.00 0.00 35.33 0.00 0.00
0.00 0.00 0.00 0.00 35.33 0.00
0.00 0.00 0.00 0.00 0.00 35.33
Bulk Modulus BV = 70.86 GPa
Shear Modulus GV = 33.46 GPa
Youngs Modulus EV = 86.74 GPa
Poission Ratio uV = 0.30
generate /gpfs01/zhq_work/fzh/work/ml/test/Al-2/test/03.vacancy/Al/std-fcc/deepmd/struct-3x3x3-000
/gpfs01/zhq_work/fzh/work/ml/test/Al-2/test/03.vacancy/Al/std-fcc/deepmd
Structure: Vac_E(eV) E(eV) equi_E(eV)
struct-3x3x3-000: 0.618 -400.452 -401.070
task poscar: /gpfs01/zhq_work/fzh/work/ml/test/Al-2/test/04.interstitial/Al/std-fcc/deepmd/POSCAR
generate /gpfs01/zhq_work/fzh/work/ml/test/Al-2/test/04.interstitial/Al/std-fcc/deepmd/struct-Al-3x3x3-000
generate /gpfs01/zhq_work/fzh/work/ml/test/Al-2/test/04.interstitial/Al/std-fcc/deepmd/struct-Al-3x3x3-001
/gpfs01/zhq_work/fzh/work/ml/test/Al-2/test/04.interstitial/Al/std-fcc/deepmd/struct-Al-3x3x3-*
Insert_ele-Struct: Inter_E(eV) E(eV) equi_E(eV)
struct-Al-3x3x3-000: 0.720 -407.846 -408.566
struct-Al-3x3x3-001: 0.835 -407.731 -408.566
Best regards,
Zhongheng