def build_groups(self,
mpsnl,
force_new=False,
snlgroup_guess=None,
testing_mode=False):
# testing mode is used to see if something already exists in DB w/o adding it to the db
match_found = False
if not force_new:
if snlgroup_guess:
sgp = self.snlgroups.find_one({'snlgroup_id': snlgroup_guess})
snlgroup = SNLGroup.from_dict(sgp)
match_found = self._add_if_belongs(snlgroup, mpsnl,
testing_mode)
if not match_found:
# look at all potential matches
for entry in self.snlgroups.find(
{'snlgroup_key': mpsnl.snlgroup_key},
sort=[("num_snl", DESCENDING)]):
snlgroup = SNLGroup.from_dict(entry)
match_found = self._add_if_belongs(snlgroup, mpsnl,
testing_mode)
if match_found:
break
if not match_found:
# add a new SNLGroup
snlgroup_id = self._get_next_snlgroup_id()
snlgroup = SNLGroup(snlgroup_id, mpsnl)
if not testing_mode:
self.snlgroups.insert(snlgroup.to_dict)
return snlgroup, not match_found
def build_groups(self, mpsnl, force_new=False, snlgroup_guess=None, testing_mode=False):
# testing mode is used to see if something already exists in DB w/o adding it to the db
match_found = False
if not force_new:
if snlgroup_guess:
sgp = self.snlgroups.find_one({'snlgroup_id': snlgroup_guess})
snlgroup = SNLGroup.from_dict(sgp)
match_found, spec_group = self._add_if_belongs(snlgroup, mpsnl, testing_mode)
if not match_found:
# look at all potential matches
for entry in self.snlgroups.find({'snlgroup_key': mpsnl.snlgroup_key},
sort=[("num_snl", DESCENDING)]):
snlgroup = SNLGroup.from_dict(entry)
match_found, spec_group = self._add_if_belongs(snlgroup, mpsnl, testing_mode)
if match_found:
break
if not match_found:
# add a new SNLGroup
snlgroup_id = self._get_next_snlgroup_id()
snlgroup = SNLGroup(snlgroup_id, mpsnl)
spec_group=None
if snlgroup.species_groups:
spec_group = snlgroup.species_groups.keys()[0]
if not testing_mode:
self.snlgroups.insert(snlgroup.to_dict)
return snlgroup, not match_found, spec_group
def switch_canonical_snl(self, snlgroup_id, canonical_mpsnl):
sgp = self.snlgroups.find_one({'snlgroup_id': snlgroup_id})
snlgroup = SNLGroup.from_dict(sgp)
all_snl_ids = [sid for sid in snlgroup.all_snl_ids]
if canonical_mpsnl.snl_id not in all_snl_ids:
raise ValueError('Canonical SNL must already be in snlgroup to switch!')
new_group = SNLGroup(snlgroup_id, canonical_mpsnl, all_snl_ids)
self.snlgroups.update({'snlgroup_id': snlgroup_id}, new_group.as_dict())
def switch_canonical_snl(self, snlgroup_id, canonical_mpsnl):
sgp = self.snlgroups.find_one({'snlgroup_id': snlgroup_id})
snlgroup = SNLGroup.from_dict(sgp)
all_snl_ids = [sid for sid in snlgroup.all_snl_ids]
if canonical_mpsnl.snl_id not in all_snl_ids:
raise ValueError(
'Canonical SNL must already be in snlgroup to switch!')
new_group = SNLGroup(snlgroup_id, canonical_mpsnl, all_snl_ids)
self.snlgroups.update({'snlgroup_id': snlgroup_id}, new_group.to_dict)
def process_item(self, item, index):
nrow, ncol = index/self._ncols, index%self._ncols
snlgroups = {} # keep {snlgroup_id: SNLGroup} to avoid dupe queries
if isinstance(item, dict) and 'snlgroup_ids' in item:
for gid in item['snlgroup_ids']:
try:
snlgrp_dict = self._snlgroups.collection.find_one({ "snlgroup_id": gid })
snlgroups[gid] = SNLGroup.from_dict(snlgrp_dict)
except:
exc_type, exc_value, exc_traceback = sys.exc_info()
_log.info('%r %r', exc_type, exc_value)
self._increase_counter(nrow, ncol, {categories[self.checker_name]: [str(gid)]})
return nrow, ncol, snlgroups
def process_item(self, item, index):
nrow, ncol = index / self._ncols, index % self._ncols
snlgroups = {} # keep {snlgroup_id: SNLGroup} to avoid dupe queries
if isinstance(item, dict) and 'snlgroup_ids' in item:
for gid in item['snlgroup_ids']:
try:
snlgrp_dict = self._snlgroups.collection.find_one(
{"snlgroup_id": gid})
snlgroups[gid] = SNLGroup.from_dict(snlgrp_dict)
except:
exc_type, exc_value, exc_traceback = sys.exc_info()
_log.info('%r %r', exc_type, exc_value)
self._increase_counter(
nrow, ncol,
{categories[self.checker_name]: [str(gid)]})
return nrow, ncol, snlgroups
def build_groups(self, mpsnl, testing_mode=False):
# testing mode is used to see if something already exists in DB w/o adding it to the db
add_new = True
for entry in self.snlgroups.find({'snlgroup_key': mpsnl.snlgroup_key},
sort=[("num_snl", DESCENDING)]):
snlgroup = SNLGroup.from_dict(entry)
if snlgroup.add_if_belongs(mpsnl):
add_new = False
print 'MATCH FOUND, grouping (snl_id, snlgroup): {}'.format(
(mpsnl.snl_id, snlgroup.snlgroup_id))
if not testing_mode:
self.snlgroups.update({'snlgroup_id': snlgroup.snlgroup_id}, snlgroup.to_dict)
break
if add_new:
# add a new SNLGroup
snlgroup_id = self._get_next_snlgroup_id()
snlgroup = SNLGroup(snlgroup_id, mpsnl)
if not testing_mode:
self.snlgroups.insert(snlgroup.to_dict)
return snlgroup, add_new
def analyze(args):
"""analyze data at any point for a copy of the streaming figure"""
# NOTE: make copy online first with suffix _%Y-%m-%d and note figure id
fig = py.get_figure(creds['username'], args.fig_id)
if args.t:
if args.fig_id == 42:
label_entries = filter(
None, '<br>'.join(fig['data'][2]['text']).split('<br>'))
pairs = map(make_tuple, label_entries)
grps = set(chain.from_iterable(pairs))
snlgrp_cursor = sma.snlgroups.aggregate([{
'$match': {
'snlgroup_id': {
'$in': list(grps)
},
'canonical_snl.about.projects': {
'$ne': 'CederDahn Challenge'
}
}
}, {
'$project': {
'snlgroup_id': 1,
'canonical_snl.snlgroup_key': 1,
'_id': 0
}
}],
cursor={})
snlgroup_keys = {}
for d in snlgrp_cursor:
snlgroup_keys[
d['snlgroup_id']] = d['canonical_snl']['snlgroup_key']
print snlgroup_keys[40890]
sma2 = SNLMongoAdapter.from_file(
os.path.join(os.environ['DB_LOC'], 'materials_db.yaml'))
materials_cursor = sma2.database.materials.aggregate([{
'$match': {
'snlgroup_id_final': {
'$in': list(grps)
},
'snl_final.about.projects': {
'$ne': 'CederDahn Challenge'
}
}
}, {
'$project': {
'snlgroup_id_final': 1,
'_id': 0,
'task_id': 1,
'final_energy_per_atom': 1,
'band_gap.search_gap.band_gap': 1,
'volume': 1,
'nsites': 1
}
}],
cursor={})
snlgroup_data = {}
for material in materials_cursor:
snlgroup_id = material['snlgroup_id_final']
final_energy_per_atom = material['final_energy_per_atom']
band_gap = material['band_gap']['search_gap']['band_gap']
volume_per_atom = material['volume'] / material['nsites']
snlgroup_data[snlgroup_id] = {
'final_energy_per_atom': final_energy_per_atom,
'band_gap': band_gap,
'task_id': material['task_id'],
'volume_per_atom': volume_per_atom
}
print snlgroup_data[40890]
filestem = 'mpworks/check_snl/results/bad_snlgroups_2_'
with open(filestem+'in_matdb.csv', 'wb') as f, \
open(filestem+'notin_matdb.csv', 'wb') as g:
writer1, writer2 = csv.writer(f), csv.writer(g)
header = [
'category', 'composition', 'snlgroup_id 1', 'sg_num 1',
'task_id 1', 'snlgroup_id 2', 'sg_num 2', 'task_id 2',
'delta_energy', 'delta_bandgap', 'delta_volume_per_atom',
'rms_dist', 'scenario'
]
writer1.writerow(header)
writer2.writerow(header)
for primary_id, secondary_id in pairs:
if primary_id not in snlgroup_keys or \
secondary_id not in snlgroup_keys:
continue
composition, primary_sg_num = snlgroup_keys[
primary_id].split('--')
secondary_sg_num = snlgroup_keys[secondary_id].split(
'--')[1]
category = 'same SGs' if primary_sg_num == secondary_sg_num else 'diff. SGs'
if primary_id not in snlgroup_data or secondary_id not in snlgroup_data:
delta_energy, delta_bandgap, delta_volume_per_atom = '', '', ''
else:
delta_energy = "{0:.3g}".format(abs(
snlgroup_data[primary_id]['final_energy_per_atom'] - \
snlgroup_data[secondary_id]['final_energy_per_atom']
))
delta_bandgap = "{0:.3g}".format(abs(
snlgroup_data[primary_id]['band_gap'] - \
snlgroup_data[secondary_id]['band_gap']
))
delta_volume_per_atom = "{0:.3g}".format(abs(
snlgroup_data[primary_id]['volume_per_atom'] - \
snlgroup_data[secondary_id]['volume_per_atom']
))
scenario, rms_dist_str = '', ''
if category == 'diff. SGs' and delta_energy and delta_bandgap:
scenario = 'different' if (
float(delta_energy) > 0.01
or float(delta_bandgap) > 0.1) else 'similar'
snlgrp1_dict = sma.snlgroups.find_one(
{"snlgroup_id": primary_id})
snlgrp2_dict = sma.snlgroups.find_one(
{"snlgroup_id": secondary_id})
snlgrp1 = SNLGroup.from_dict(snlgrp1_dict)
snlgrp2 = SNLGroup.from_dict(snlgrp2_dict)
primary_structure = snlgrp1.canonical_structure
secondary_structure = snlgrp2.canonical_structure
rms_dist = matcher.get_rms_dist(
primary_structure, secondary_structure)
if rms_dist is not None:
rms_dist_str = "({0:.3g},{1:.3g})".format(
*rms_dist)
print rms_dist_str
row = [
category, composition,
primary_id, primary_sg_num,
snlgroup_data[primary_id]['task_id'] \
if primary_id in snlgroup_data else '',
secondary_id, secondary_sg_num,
snlgroup_data[secondary_id]['task_id'] \
if secondary_id in snlgroup_data else '',
delta_energy, delta_bandgap, delta_volume_per_atom,
rms_dist_str, scenario
]
if delta_energy and delta_bandgap: writer1.writerow(row)
else: writer2.writerow(row)
elif args.fig_id == 16:
out_fig = Figure()
badsnls_trace = Scatter(x=[],
y=[],
text=[],
mode='markers',
name='SG Changes')
bisectrix = Scatter(x=[0, 230],
y=[0, 230],
mode='lines',
name='bisectrix')
print 'pulling bad snls from plotly ...'
bad_snls = OrderedDict()
for category, text in zip(fig['data'][2]['y'],
fig['data'][2]['text']):
for snl_id in map(int, text.split('<br>')):
bad_snls[snl_id] = category
with open('mpworks/check_snl/results/bad_snls.csv', 'wb') as f:
print 'pulling bad snls from database ...'
mpsnl_cursor = sma.snl.find({
'snl_id': {
'$in': bad_snls.keys()
},
'about.projects': {
'$ne': 'CederDahn Challenge'
}
})
writer = csv.writer(f)
writer.writerow([
'snl_id', 'category', 'snlgroup_key', 'nsites', 'remarks',
'projects', 'authors'
])
print 'writing bad snls to file ...'
for mpsnl_dict in mpsnl_cursor:
mpsnl = MPStructureNL.from_dict(mpsnl_dict)
row = [
mpsnl.snl_id, bad_snls[mpsnl.snl_id],
mpsnl.snlgroup_key
]
row += _get_snl_extra_info(mpsnl)
writer.writerow(row)
sg_num = mpsnl.snlgroup_key.split('--')[1]
if (bad_snls[mpsnl.snl_id] == 'SG default' and sg_num != '-1') or \
bad_snls[mpsnl.snl_id] == 'SG change':
mpsnl.structure.remove_oxidation_states()
sf = SpacegroupAnalyzer(mpsnl.structure, symprec=0.1)
badsnls_trace['x'].append(mpsnl.sg_num)
badsnls_trace['y'].append(sf.get_spacegroup_number())
badsnls_trace['text'].append(mpsnl.snl_id)
if bad_snls[mpsnl.snl_id] == 'SG default':
print sg_num, sf.get_spacegroup_number()
print 'plotting out-fig ...'
out_fig['data'] = Data([bisectrix, badsnls_trace])
out_fig['layout'] = Layout(
showlegend=False,
hovermode='closest',
title='Spacegroup Assignment Changes',
xaxis=XAxis(showgrid=False,
title='old SG number',
range=[0, 230]),
yaxis=YAxis(showgrid=False,
title='new SG number',
range=[0, 230]),
)
filename = 'spacegroup_changes_'
filename += datetime.datetime.now().strftime('%Y-%m-%d')
py.plot(out_fig, filename=filename, auto_open=False)
elif args.fig_id == 43: # SNLGroupMemberChecker
matcher2 = StructureMatcher(ltol=0.2,
stol=0.3,
angle_tol=5,
primitive_cell=False,
scale=True,
attempt_supercell=True,
comparator=ElementComparator())
print 'pulling data from plotly ...'
trace = Scatter(x=[],
y=[],
text=[],
mode='markers',
name='mismatches')
bad_snls = OrderedDict() # snlgroup_id : [ mismatching snl_ids ]
for category, text in zip(fig['data'][2]['y'],
fig['data'][2]['text']):
if category != 'mismatch': continue
for entry in text.split('<br>'):
fields = entry.split(':')
snlgroup_id = int(fields[0].split(',')[0])
print snlgroup_id
snlgrp_dict = sma.snlgroups.find_one(
{'snlgroup_id': snlgroup_id})
snlgrp = SNLGroup.from_dict(snlgrp_dict)
s1 = snlgrp.canonical_structure.get_primitive_structure()
bad_snls[snlgroup_id] = []
for i, snl_id in enumerate(fields[1].split(',')):
mpsnl_dict = sma.snl.find_one({'snl_id': int(snl_id)})
if 'CederDahn Challenge' in mpsnl_dict['about'][
'projects']:
print 'skip CederDahn: %s' % snl_id
continue
mpsnl = MPStructureNL.from_dict(mpsnl_dict)
s2 = mpsnl.structure.get_primitive_structure()
is_match = matcher2.fit(s1, s2)
if is_match: continue
bad_snls[snlgroup_id].append(snl_id)
trace['x'].append(snlgroup_id)
trace['y'].append(i + 1)
trace['text'].append(snl_id)
if len(bad_snls[snlgroup_id]) < 1:
bad_snls.pop(snlgroup_id, None)
with open('mpworks/check_snl/results/bad_snlgroups.csv',
'wb') as f:
print 'pulling bad snlgroups from database ...'
snlgroup_cursor = sma.snlgroups.find({
'snlgroup_id': {
'$in': bad_snls.keys()
},
})
writer = csv.writer(f)
writer.writerow(
['snlgroup_id', 'snlgroup_key', 'mismatching snl_ids'])
print 'writing bad snlgroups to file ...'
for snlgroup_dict in snlgroup_cursor:
snlgroup = SNLGroup.from_dict(snlgroup_dict)
row = [
snlgroup.snlgroup_id,
snlgroup.canonical_snl.snlgroup_key,
' '.join(bad_snls[snlgroup.snlgroup_id])
]
writer.writerow(row)
print 'plotting out-fig ...'
out_fig = Figure()
out_fig['data'] = Data([trace])
out_fig['layout'] = Layout(
showlegend=False,
hovermode='closest',
title='Member Mismatches of SNLGroup Canonicals',
xaxis=XAxis(showgrid=False,
title='snlgroup_id',
showexponent='none'),
yaxis=YAxis(showgrid=False, title='# mismatching SNLs'),
)
filename = 'groupmember_mismatches_'
filename += datetime.datetime.now().strftime('%Y-%m-%d')
py.plot(out_fig, filename=filename, auto_open=False)
else:
errors = Counter()
bad_snls = OrderedDict()
bad_snlgroups = OrderedDict()
for i, d in enumerate(fig['data']):
if not isinstance(d, Scatter): continue
if not 'x' in d or not 'y' in d or not 'text' in d: continue
start_id = int(d['name'].split(' - ')[0][:-1]) * 1000
marker_colors = d['marker']['color']
if i < 2 * num_snl_streams: # spacegroups
errors += Counter(marker_colors)
for idx, color in enumerate(marker_colors):
snl_id = start_id + d['x'][idx]
color_index = category_colors.index(color)
category = categories[color_index]
bad_snls[snl_id] = category
else: # groupmembers
for idx, color in enumerate(marker_colors):
if color != category_colors[0]: continue
snlgroup_id = start_id + d['x'][idx]
mismatch_snl_id, canonical_snl_id = d['text'][idx].split(
' != ')
bad_snlgroups[snlgroup_id] = int(mismatch_snl_id)
print errors
fig_data = fig['data'][-1]
fig_data['x'] = [
errors[color] for color in fig_data['marker']['color']
]
filename = _get_filename()
print filename
#py.plot(fig, filename=filename)
with open('mpworks/check_snl/results/bad_snls.csv', 'wb') as f:
mpsnl_cursor = sma.snl.find({'snl_id': {'$in': bad_snls.keys()}})
writer = csv.writer(f)
writer.writerow([
'snl_id', 'category', 'snlgroup_key', 'nsites', 'remarks',
'projects', 'authors'
])
for mpsnl_dict in mpsnl_cursor:
mpsnl = MPStructureNL.from_dict(mpsnl_dict)
row = [
mpsnl.snl_id, bad_snls[mpsnl.snl_id], mpsnl.snlgroup_key
]
row += _get_snl_extra_info(mpsnl)
writer.writerow(row)
with open('mpworks/check_snl/results/bad_snlgroups.csv', 'wb') as f:
snlgrp_cursor = sma.snlgroups.find(
{'snlgroup_id': {
'$in': bad_snlgroups.keys()
}})
first_mismatch_snls_cursor = sma.snl.find(
{'snl_id': {
'$in': bad_snlgroups.values()
}})
first_mismatch_snl_info = OrderedDict()
for mpsnl_dict in first_mismatch_snls_cursor:
mpsnl = MPStructureNL.from_dict(mpsnl_dict)
first_mismatch_snl_info[mpsnl.snl_id] = _get_snl_extra_info(
mpsnl)
writer = csv.writer(f)
writer.writerow([
'snlgroup_id', 'snlgroup_key', 'canonical_snl_id',
'first_mismatching_snl_id', 'nsites', 'remarks', 'projects',
'authors'
])
for snlgrp_dict in snlgrp_cursor:
snlgrp = SNLGroup.from_dict(snlgrp_dict)
first_mismatch_snl_id = bad_snlgroups[snlgrp.snlgroup_id]
row = [
snlgrp.snlgroup_id, snlgrp.canonical_snl.snlgroup_key,
snlgrp.canonical_snl.snl_id, first_mismatch_snl_id
]
row += [
' & '.join(pair) if pair[0] != pair[1] else pair[0]
for pair in zip(
_get_snl_extra_info(snlgrp.canonical_snl),
first_mismatch_snl_info[int(first_mismatch_snl_id)])
]
writer.writerow(row)
def check_snls_in_snlgroups(args):
"""check whether SNLs in each SNLGroup still match resp. canonical SNL"""
range_index = args.start / num_ids_per_stream
idxs = [2 * (num_snl_streams + range_index)]
idxs += [idxs[0] + 1]
s = [py.Stream(stream_ids[i]) for i in idxs]
for i in range(len(idxs)):
s[i].open()
end = num_snlgroups if args.end > num_snlgroups else args.end
id_range = {"$gt": args.start, "$lte": end}
snlgrp_cursor = sma.snlgroups.find({"snlgroup_id": id_range})
colors = []
num_good_ids = 0
for snlgrp_dict in snlgrp_cursor:
start_time = time.clock()
try:
snlgrp = SNLGroup.from_dict(snlgrp_dict)
except:
exc_type, exc_value, exc_traceback = sys.exc_info()
text = ' '.join([str(exc_type), str(exc_value)])
colors.append(category_colors[-1]) # Others
data = dict(x=snlgrp_dict['snlgroup_id'] % num_ids_per_stream,
y=range_index,
text=text,
marker=Marker(color=colors))
s[0].write(data)
sleep(start_time)
continue
if len(snlgrp.all_snl_ids) <= 1:
num_good_ids += 1
data = dict(x=[num_good_ids], y=[range_index])
s[1].write(data)
sleep(start_time)
continue
exc_raised = False
all_snls_good = True
for snl_id in snlgrp.all_snl_ids:
if snl_id == snlgrp.canonical_snl.snl_id: continue
mpsnl_dict = sma.snl.find_one({"snl_id": snl_id})
try:
mpsnl = MPStructureNL.from_dict(mpsnl_dict)
is_match = matcher.fit(mpsnl.structure,
snlgrp.canonical_structure)
except:
exc_type, exc_value, exc_traceback = sys.exc_info()
exc_raised = True
if exc_raised or not is_match: # Scatter (bad)
if exc_raised:
category = 2 if fnmatch(str(exc_type), '*pybtex*') else 3
text = ' '.join([str(exc_type), str(exc_value)])
else:
category = 0
text = '%d != can:%d' % (mpsnl_dict['snl_id'],
snlgrp.canonical_snl.snl_id)
colors.append(category_colors[category])
data = dict(x=snlgrp_dict['snlgroup_id'] % num_ids_per_stream,
y=range_index,
text=text,
marker=Marker(color=colors))
s[0].write(data)
all_snls_good = False
sleep(start_time)
break
if all_snls_good: # Bar (good)
num_good_ids += 1
data = dict(x=[num_good_ids], y=[range_index])
s[1].write(data)
sleep(start_time)
for i in range(len(idxs)):
s[i].close()
def analyze(args):
"""analyze data at any point for a copy of the streaming figure"""
# NOTE: make copy online first with suffix _%Y-%m-%d and note figure id
fig = py.get_figure(creds['username'], args.fig_id)
if args.t:
if args.fig_id == 42:
label_entries = filter(None, '<br>'.join(fig['data'][2]['text']).split('<br>'))
pairs = map(make_tuple, label_entries)
grps = set(chain.from_iterable(pairs))
snlgrp_cursor = sma.snlgroups.aggregate([
{ '$match': {
'snlgroup_id': { '$in': list(grps) },
'canonical_snl.about.projects': {'$ne': 'CederDahn Challenge'}
} },
{ '$project': { 'snlgroup_id': 1, 'canonical_snl.snlgroup_key': 1, '_id': 0 } }
], cursor={})
snlgroup_keys = {}
for d in snlgrp_cursor:
snlgroup_keys[d['snlgroup_id']] = d['canonical_snl']['snlgroup_key']
print snlgroup_keys[40890]
sma2 = SNLMongoAdapter.from_file(
os.path.join(os.environ['DB_LOC'], 'materials_db.yaml')
)
materials_cursor = sma2.database.materials.aggregate([
{ '$match': {
'snlgroup_id_final': { '$in': list(grps) },
'snl_final.about.projects': {'$ne': 'CederDahn Challenge'}
} },
{ '$project': {
'snlgroup_id_final': 1, '_id': 0, 'task_id': 1,
'final_energy_per_atom': 1,
'band_gap.search_gap.band_gap': 1,
'volume': 1, 'nsites': 1
}}
], cursor={})
snlgroup_data = {}
for material in materials_cursor:
snlgroup_id = material['snlgroup_id_final']
final_energy_per_atom = material['final_energy_per_atom']
band_gap = material['band_gap']['search_gap']['band_gap']
volume_per_atom = material['volume'] / material['nsites']
snlgroup_data[snlgroup_id] = {
'final_energy_per_atom': final_energy_per_atom,
'band_gap': band_gap, 'task_id': material['task_id'],
'volume_per_atom': volume_per_atom
}
print snlgroup_data[40890]
filestem = 'mpworks/check_snl/results/bad_snlgroups_2_'
with open(filestem+'in_matdb.csv', 'wb') as f, \
open(filestem+'notin_matdb.csv', 'wb') as g:
writer1, writer2 = csv.writer(f), csv.writer(g)
header = [
'category', 'composition',
'snlgroup_id 1', 'sg_num 1', 'task_id 1',
'snlgroup_id 2', 'sg_num 2', 'task_id 2',
'delta_energy', 'delta_bandgap', 'delta_volume_per_atom',
'rms_dist', 'scenario'
]
writer1.writerow(header)
writer2.writerow(header)
for primary_id, secondary_id in pairs:
if primary_id not in snlgroup_keys or \
secondary_id not in snlgroup_keys: continue
composition, primary_sg_num = snlgroup_keys[primary_id].split('--')
secondary_sg_num = snlgroup_keys[secondary_id].split('--')[1]
category = 'same SGs' if primary_sg_num == secondary_sg_num else 'diff. SGs'
if primary_id not in snlgroup_data or secondary_id not in snlgroup_data:
delta_energy, delta_bandgap, delta_volume_per_atom = '', '', ''
else:
delta_energy = "{0:.3g}".format(abs(
snlgroup_data[primary_id]['final_energy_per_atom'] - \
snlgroup_data[secondary_id]['final_energy_per_atom']
))
delta_bandgap = "{0:.3g}".format(abs(
snlgroup_data[primary_id]['band_gap'] - \
snlgroup_data[secondary_id]['band_gap']
))
delta_volume_per_atom = "{0:.3g}".format(abs(
snlgroup_data[primary_id]['volume_per_atom'] - \
snlgroup_data[secondary_id]['volume_per_atom']
))
scenario, rms_dist_str = '', ''
if category == 'diff. SGs' and delta_energy and delta_bandgap:
scenario = 'different' if (
float(delta_energy) > 0.01 or float(delta_bandgap) > 0.1
) else 'similar'
snlgrp1_dict = sma.snlgroups.find_one({ "snlgroup_id": primary_id })
snlgrp2_dict = sma.snlgroups.find_one({ "snlgroup_id": secondary_id })
snlgrp1 = SNLGroup.from_dict(snlgrp1_dict)
snlgrp2 = SNLGroup.from_dict(snlgrp2_dict)
primary_structure = snlgrp1.canonical_structure
secondary_structure = snlgrp2.canonical_structure
rms_dist = matcher.get_rms_dist(primary_structure, secondary_structure)
if rms_dist is not None:
rms_dist_str = "({0:.3g},{1:.3g})".format(*rms_dist)
print rms_dist_str
row = [
category, composition,
primary_id, primary_sg_num,
snlgroup_data[primary_id]['task_id'] \
if primary_id in snlgroup_data else '',
secondary_id, secondary_sg_num,
snlgroup_data[secondary_id]['task_id'] \
if secondary_id in snlgroup_data else '',
delta_energy, delta_bandgap, delta_volume_per_atom,
rms_dist_str, scenario
]
if delta_energy and delta_bandgap: writer1.writerow(row)
else: writer2.writerow(row)
elif args.fig_id == 16:
out_fig = Figure()
badsnls_trace = Scatter(x=[], y=[], text=[], mode='markers', name='SG Changes')
bisectrix = Scatter(x=[0,230], y=[0,230], mode='lines', name='bisectrix')
print 'pulling bad snls from plotly ...'
bad_snls = OrderedDict()
for category, text in zip(fig['data'][2]['y'], fig['data'][2]['text']):
for snl_id in map(int, text.split('<br>')):
bad_snls[snl_id] = category
with open('mpworks/check_snl/results/bad_snls.csv', 'wb') as f:
print 'pulling bad snls from database ...'
mpsnl_cursor = sma.snl.find({
'snl_id': { '$in': bad_snls.keys() },
'about.projects': {'$ne': 'CederDahn Challenge'}
})
writer = csv.writer(f)
writer.writerow([
'snl_id', 'category', 'snlgroup_key', 'nsites', 'remarks', 'projects', 'authors'
])
print 'writing bad snls to file ...'
for mpsnl_dict in mpsnl_cursor:
mpsnl = MPStructureNL.from_dict(mpsnl_dict)
row = [ mpsnl.snl_id, bad_snls[mpsnl.snl_id], mpsnl.snlgroup_key ]
row += _get_snl_extra_info(mpsnl)
writer.writerow(row)
sg_num = mpsnl.snlgroup_key.split('--')[1]
if (bad_snls[mpsnl.snl_id] == 'SG default' and sg_num != '-1') or \
bad_snls[mpsnl.snl_id] == 'SG change':
mpsnl.structure.remove_oxidation_states()
sf = SpacegroupAnalyzer(mpsnl.structure, symprec=0.1)
badsnls_trace['x'].append(mpsnl.sg_num)
badsnls_trace['y'].append(sf.get_spacegroup_number())
badsnls_trace['text'].append(mpsnl.snl_id)
if bad_snls[mpsnl.snl_id] == 'SG default':
print sg_num, sf.get_spacegroup_number()
print 'plotting out-fig ...'
out_fig['data'] = Data([bisectrix, badsnls_trace])
out_fig['layout'] = Layout(
showlegend=False, hovermode='closest',
title='Spacegroup Assignment Changes',
xaxis=XAxis(showgrid=False, title='old SG number', range=[0,230]),
yaxis=YAxis(showgrid=False, title='new SG number', range=[0,230]),
)
filename = 'spacegroup_changes_'
filename += datetime.datetime.now().strftime('%Y-%m-%d')
py.plot(out_fig, filename=filename, auto_open=False)
elif args.fig_id == 43: # SNLGroupMemberChecker
matcher2 = StructureMatcher(
ltol=0.2, stol=0.3, angle_tol=5, primitive_cell=False, scale=True,
attempt_supercell=True, comparator=ElementComparator()
)
print 'pulling data from plotly ...'
trace = Scatter(x=[], y=[], text=[], mode='markers', name='mismatches')
bad_snls = OrderedDict() # snlgroup_id : [ mismatching snl_ids ]
for category, text in zip(fig['data'][2]['y'], fig['data'][2]['text']):
if category != 'mismatch': continue
for entry in text.split('<br>'):
fields = entry.split(':')
snlgroup_id = int(fields[0].split(',')[0])
print snlgroup_id
snlgrp_dict = sma.snlgroups.find_one({ 'snlgroup_id': snlgroup_id })
snlgrp = SNLGroup.from_dict(snlgrp_dict)
s1 = snlgrp.canonical_structure.get_primitive_structure()
bad_snls[snlgroup_id] = []
for i, snl_id in enumerate(fields[1].split(',')):
mpsnl_dict = sma.snl.find_one({ 'snl_id': int(snl_id) })
if 'CederDahn Challenge' in mpsnl_dict['about']['projects']:
print 'skip CederDahn: %s' % snl_id
continue
mpsnl = MPStructureNL.from_dict(mpsnl_dict)
s2 = mpsnl.structure.get_primitive_structure()
is_match = matcher2.fit(s1, s2)
if is_match: continue
bad_snls[snlgroup_id].append(snl_id)
trace['x'].append(snlgroup_id)
trace['y'].append(i+1)
trace['text'].append(snl_id)
if len(bad_snls[snlgroup_id]) < 1:
bad_snls.pop(snlgroup_id, None)
with open('mpworks/check_snl/results/bad_snlgroups.csv', 'wb') as f:
print 'pulling bad snlgroups from database ...'
snlgroup_cursor = sma.snlgroups.find({
'snlgroup_id': { '$in': bad_snls.keys() },
})
writer = csv.writer(f)
writer.writerow(['snlgroup_id', 'snlgroup_key', 'mismatching snl_ids'])
print 'writing bad snlgroups to file ...'
for snlgroup_dict in snlgroup_cursor:
snlgroup = SNLGroup.from_dict(snlgroup_dict)
row = [
snlgroup.snlgroup_id, snlgroup.canonical_snl.snlgroup_key,
' '.join(bad_snls[snlgroup.snlgroup_id])
]
writer.writerow(row)
print 'plotting out-fig ...'
out_fig = Figure()
out_fig['data'] = Data([trace])
out_fig['layout'] = Layout(
showlegend=False, hovermode='closest',
title='Member Mismatches of SNLGroup Canonicals',
xaxis=XAxis(showgrid=False, title='snlgroup_id', showexponent='none'),
yaxis=YAxis(showgrid=False, title='# mismatching SNLs'),
)
filename = 'groupmember_mismatches_'
filename += datetime.datetime.now().strftime('%Y-%m-%d')
py.plot(out_fig, filename=filename, auto_open=False)
else:
errors = Counter()
bad_snls = OrderedDict()
bad_snlgroups = OrderedDict()
for i,d in enumerate(fig['data']):
if not isinstance(d, Scatter): continue
if not 'x' in d or not 'y' in d or not 'text' in d: continue
start_id = int(d['name'].split(' - ')[0][:-1])*1000
marker_colors = d['marker']['color']
if i < 2*num_snl_streams: # spacegroups
errors += Counter(marker_colors)
for idx,color in enumerate(marker_colors):
snl_id = start_id + d['x'][idx]
color_index = category_colors.index(color)
category = categories[color_index]
bad_snls[snl_id] = category
else: # groupmembers
for idx,color in enumerate(marker_colors):
if color != category_colors[0]: continue
snlgroup_id = start_id + d['x'][idx]
mismatch_snl_id, canonical_snl_id = d['text'][idx].split(' != ')
bad_snlgroups[snlgroup_id] = int(mismatch_snl_id)
print errors
fig_data = fig['data'][-1]
fig_data['x'] = [ errors[color] for color in fig_data['marker']['color'] ]
filename = _get_filename()
print filename
#py.plot(fig, filename=filename)
with open('mpworks/check_snl/results/bad_snls.csv', 'wb') as f:
mpsnl_cursor = sma.snl.find({ 'snl_id': { '$in': bad_snls.keys() } })
writer = csv.writer(f)
writer.writerow([
'snl_id', 'category', 'snlgroup_key', 'nsites', 'remarks', 'projects', 'authors'
])
for mpsnl_dict in mpsnl_cursor:
mpsnl = MPStructureNL.from_dict(mpsnl_dict)
row = [ mpsnl.snl_id, bad_snls[mpsnl.snl_id], mpsnl.snlgroup_key ]
row += _get_snl_extra_info(mpsnl)
writer.writerow(row)
with open('mpworks/check_snl/results/bad_snlgroups.csv', 'wb') as f:
snlgrp_cursor = sma.snlgroups.find({ 'snlgroup_id': { '$in': bad_snlgroups.keys() } })
first_mismatch_snls_cursor = sma.snl.find({ 'snl_id': { '$in': bad_snlgroups.values() } })
first_mismatch_snl_info = OrderedDict()
for mpsnl_dict in first_mismatch_snls_cursor:
mpsnl = MPStructureNL.from_dict(mpsnl_dict)
first_mismatch_snl_info[mpsnl.snl_id] = _get_snl_extra_info(mpsnl)
writer = csv.writer(f)
writer.writerow([
'snlgroup_id', 'snlgroup_key',
'canonical_snl_id', 'first_mismatching_snl_id',
'nsites', 'remarks', 'projects', 'authors'
])
for snlgrp_dict in snlgrp_cursor:
snlgrp = SNLGroup.from_dict(snlgrp_dict)
first_mismatch_snl_id = bad_snlgroups[snlgrp.snlgroup_id]
row = [
snlgrp.snlgroup_id, snlgrp.canonical_snl.snlgroup_key,
snlgrp.canonical_snl.snl_id, first_mismatch_snl_id
]
row += [
' & '.join(pair) if pair[0] != pair[1] else pair[0]
for pair in zip(
_get_snl_extra_info(snlgrp.canonical_snl),
first_mismatch_snl_info[int(first_mismatch_snl_id)]
)
]
writer.writerow(row)
def check_snls_in_snlgroups(args):
"""check whether SNLs in each SNLGroup still match resp. canonical SNL"""
range_index = args.start / num_ids_per_stream
idxs = [2*(num_snl_streams+range_index)]
idxs += [idxs[0]+1]
s = [py.Stream(stream_ids[i]) for i in idxs]
for i in range(len(idxs)): s[i].open()
end = num_snlgroups if args.end > num_snlgroups else args.end
id_range = {"$gt": args.start, "$lte": end}
snlgrp_cursor = sma.snlgroups.find({ "snlgroup_id": id_range})
colors = []
num_good_ids = 0
for snlgrp_dict in snlgrp_cursor:
start_time = time.clock()
try:
snlgrp = SNLGroup.from_dict(snlgrp_dict)
except:
exc_type, exc_value, exc_traceback = sys.exc_info()
text = ' '.join([str(exc_type), str(exc_value)])
colors.append(category_colors[-1]) # Others
data = dict(
x=snlgrp_dict['snlgroup_id']%num_ids_per_stream,
y=range_index, text=text, marker=Marker(color=colors)
)
s[0].write(data)
sleep(start_time)
continue
if len(snlgrp.all_snl_ids) <= 1:
num_good_ids += 1
data = dict(x=[num_good_ids], y=[range_index])
s[1].write(data)
sleep(start_time)
continue
exc_raised = False
all_snls_good = True
for snl_id in snlgrp.all_snl_ids:
if snl_id == snlgrp.canonical_snl.snl_id: continue
mpsnl_dict = sma.snl.find_one({ "snl_id": snl_id })
try:
mpsnl = MPStructureNL.from_dict(mpsnl_dict)
is_match = matcher.fit(mpsnl.structure, snlgrp.canonical_structure)
except:
exc_type, exc_value, exc_traceback = sys.exc_info()
exc_raised = True
if exc_raised or not is_match: # Scatter (bad)
if exc_raised:
category = 2 if fnmatch(str(exc_type), '*pybtex*') else 3
text = ' '.join([str(exc_type), str(exc_value)])
else:
category = 0
text = '%d != can:%d' % (mpsnl_dict['snl_id'], snlgrp.canonical_snl.snl_id)
colors.append(category_colors[category])
data = dict(
x=snlgrp_dict['snlgroup_id']%num_ids_per_stream,
y=range_index, text=text, marker=Marker(color=colors)
)
s[0].write(data)
all_snls_good = False
sleep(start_time)
break
if all_snls_good: # Bar (good)
num_good_ids += 1
data = dict(x=[num_good_ids], y=[range_index])
s[1].write(data)
sleep(start_time)
for i in range(len(idxs)): s[i].close()