def build_deepmd(path, nsw, outcar, deepmd):
ls = LabeledSystem(outcar, fmt='outcar')
"""
sub_ls = ls.sub_system(idx)
"""
if args.idx:
print("index file provided")
idx = np.loadtxt(args.idx).astype(int)
# ls = ls.sub_system(idx)
if args.vaspidx:
print("vasp index file provided")
vaspidx = np.loadtxt(args.vaspidx)
fp = open(outcar)
fp.readline()
nsw_sel = fp.readline()
if 'nsw_sel' in nsw_sel:
print('file generated by merge_out.py')
# print(nsw_sel)
tmp = nsw_sel.split('=')[1].strip().split(' ')
nsw_sel = [int(tmp_idx) for tmp_idx in tmp]
idx = []
for i in range(len(nsw_sel)):
if nsw_sel[i] in vaspidx:
idx.append(i)
else:
print('OUTCAR file generated by VASP')
idx = vaspidx - 1
idx2 = [i for i in range(len(ls)) if i not in idx]
ls2 = ls.sub_system(idx2)
ls = ls.sub_system(idx)
deepmd = os.path.join(path, deepmd)
if args.batchsize:
set_size = args.batchsize
else:
if nsw <= 4: # we know nsw must > 100
set_size = 1
print("{0} has only {1}".format(path, nsw))
if nsw > 4:
set_size, _ = best_size(
nsw
) # 25% used as , but if say 82, then 20, 20, 20, 2, too less
ls.to_deepmd_npy(deepmd, set_size=set_size)
if args.test:
ls2.to_deepmd_npy('test_tmp', set_size=100000)
shutil.copytree('test_tmp/set.000', os.path.join(deepmd, 'set.001'))
shutil.rmtree('test_tmp')
def to_system(self, data, **kwargs):
"""
convert system to list, usefull for data collection
"""
from dpdata import System, LabeledSystem
if 'forces' in data:
system = LabeledSystem(data=data)
else:
system = System(data=data)
if len(system) == 0:
return []
if len(system) == 1:
return [system]
else:
systems = []
for ii in range(len(system)):
systems.append(system.sub_system([ii]))
return systems
def build_deepmd_frames(path, outcar, deepmd):
"""
sub_ls = ls.sub_system(idx)
"""
try:
ls = LabeledSystem(outcar, fmt=args.format)
except:
ls = System(outcar, fmt=args.format)
if args.exclude:
oldsize = len(ls)
idx_new = [i for i in range(len(ls)) if i not in args.exclude]
ls = ls.sub_system(idx_new)
newsize = len(ls)
print('{0}/{1} is selected'.format(newsize, oldsize))
if args.force_limit:
fmin = min(args.force_limit)
fmax = max(args.force_limit)
print("force limit imposed, force in between {0}, {1}".format(
fmin, fmax))
idx_new = []
exclude = []
for i in range(len(ls)):
forces = ls[i].data['forces']
if forces.min() >= fmin and forces.max() <= fmax:
idx_new.append(i)
else:
exclude.append(i)
print('excluded frames', exclude)
print('{0} / {1} is selected'.format(len(idx_new), len(ls)))
ls = ls.sub_system(idx_new)
if args.idx:
print("index file provided")
idx = np.loadtxt(args.idx).astype(int)
elif (not args.idx) and args.vaspidx:
print("vasp index file provided")
vaspidx = np.loadtxt(args.vaspidx)
fp = open(outcar)
fp.readline()
nsw_sel = fp.readline()
if 'nsw_sel' in nsw_sel:
print('file generated by merge_out.py')
tmp = nsw_sel.split('=')[1].strip().split(' ')
nsw_sel = [int(tmp_idx) for tmp_idx in tmp]
idx = []
for i in range(len(nsw_sel)):
if nsw_sel[i] in vaspidx:
idx.append(i)
else:
print('OUTCAR file generated by VASP')
idx = vaspidx - 1
else:
print("split train and test by ratio {0} : {1}".format(
args.train_test_ratio, 1))
train_size = round(
len(ls) * (args.train_test_ratio) / (args.train_test_ratio + 1))
idx = np.random.choice(range(len(ls)), train_size, replace=False)
idx.sort()
idx2 = [i for i in range(len(ls)) if i not in idx] # test
ls2 = ls.sub_system(idx2) # test
ls = ls.sub_system(idx)
deepmd = os.path.join(path, deepmd)
ls.to_deepmd_npy(deepmd,
set_size=1000000) # give a *large* value, default is 5000
if len(ls2) == 0:
print('test set has no data')
elif args.savetest and len(ls2) > 0:
ls2.to_deepmd_npy('test_tmp', set_size=1000000)
shutil.copytree('test_tmp/set.000', os.path.join(deepmd, 'set.001'))
shutil.rmtree('test_tmp')
#max_f = []
#for f in force:
# max_f.append(np.max(np.max(f)))
#print(max_f)
#plt.plot(max_f)
tmp = '/Users/jiedeng/GD/papers/pv3_crystallization/post_nn/exsolution_pert/u.project.ESS.lstixrud.jd848.pv+hf.dp-train.lmp_run.6k.rp5.160-cpu.pert.10k_good_p3.recal/deepmd_all'
tmp2 = '/Users/jiedeng/GD/papers/pv3_crystallization/post_nn/exsolution_pert/u.home.j.jd848.project-lstixrud.metad.3rd.recal/deepmd/'
ls = LabeledSystem(tmp2,fmt='deepmd/npy')
from dpdata import LabeledSystem
import numpy as np
ls = LabeledSystem('deepmd_ttr2',fmt='deepmd/npy')
idx = list(range(len(ls)))
idx.remove(14)
ls2 = ls.sub_system(idx)
ls2.to_deepmd_npy('test')
fparam = np.load('deepmd_ttr2/set.001/fparam.npy')
np.save('test/set.000/fparam.npy',fparam[idx])
cp -r deepmd_ttr2 deepmd_ttr2_tmp
rm -r deepmd_ttr2_tmp/set.000
from dpdata import LabeledSystem
import numpy as np
ls = LabeledSystem('deepmd_ttr2_tmp',fmt='deepmd/npy')
idx = list(range(len(ls)))
idx.remove(10)
ls2 = ls.sub_system(idx)
ls2.to_deepmd_npy('test')
fparam = np.load('deepmd_ttr2_tmp/set.000/fparam.npy')
from glob import glob
from tqdm import tqdm
"""
process multi systems
"""
fs = glob('iter.0000[4-7]*/02.fp/task*/OUTCAR')
maxf = 3.0
ms = MultiSystems()
ic = 0
vacuum_size = 13
for f in tqdm(fs):
if check_cluster(f.replace('OUTCAR', 'POSCAR'), vacuum_size, fmt='POSCAR'):
print(f)
continue
try:
ls = LabeledSystem(f)
except:
print(f)
continue
if len(ls) > 0:
if ls.sub_system([0]).data['forces'].max() > maxf:
pass
else:
ic += 1
ms.append(ls)
print(len(fs))
print(ic)
ms.to_deepmd_raw('deepmd-f%s' % maxf)
ms.to_deepmd_npy('deepmd-f%s' % maxf)
from glob import glob
from dpdata import LabeledSystem
from monty.serialization import dumpfn, loadfn
from tqdm import tqdm
fs = glob('usefull-[1-3]/sys-*/OUTCAR')
entries = []
for f in tqdm(fs):
ls = LabeledSystem(f)
ls.sub_system([-1]).to_pymatgen_ComputedStructureEntry()
entry = ls.sub_system([-1]).to_pymatgen_ComputedStructureEntry()[0]
entries.append(entry)
dumpfn(entries, 'all-vasp-entries.json')
