def build_deepmd(path,nsw):
ls = LabeledSystem(os.path.join(path, 'outcar'),fmt='outcar')
deepmd = os.path.join(path,'deepmd')
if nsw <= 4: # we know nsw must > 100
set_size = 1
print("{0} has only {1}".format(path,nsw))
if nsw > 4:
set_size = nsw//4 # 25% used as , but if say 82, then 20, 20, 20, 2, too less
ls.to_deepmd_npy(deepmd,set_size=set_size)
def build_deepmd(path,nsw):
ls=LabeledSystem(os.path.join(path, 'OUTCAR'),fmt='outcar')
deepmd = os.path.join(path,'deepmd')
if nsw <= 2000: # we know nsw must > 100
set_size = nsw//2
if nsw > 2000:
set_size = 1000
ls.to_deepmd_npy(deepmd,set_size=set_size)
if nsw>3000:
check_sets(deepmd)
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 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')
from dpdata import System, LabeledSystem, MultiSystems
import os
fp = open('folders_to_merge', 'r')
folders_org = fp.readlines()
folders = []
fp.close()
for i in range(len(folders_org)):
if '#' in folders_org[i] or folders_org[i] == '\n':
pass
else:
folders.append(folders_org[i].replace('\n', ''))
for path in folders:
pwd = os.getcwd()
os.chdir(path)
print("process ", path)
#s=System('POSCAR',fmt='poscar')
ls = LabeledSystem('OUTCAR', fmt='outcar')
ls.to_deepmd_raw('deepmd')
ls.to_deepmd_npy('deepmd', set_size=1000)
os.chdir(pwd)
print("done ", path)
print("done")
