def get_band_center(form):
c = Composition(str(form))
prod = 1.0
for el, amt in c.get_el_amt_dict().iteritems():
prod = prod * (Element(el).X ** amt)
return -prod ** (1 / sum(c.get_el_amt_dict().values()))
def _parse_criteria(self, criteria):
"""
Internal method to perform mapping of criteria to proper mongo queries
using aliases, as well as some useful sanitization. For example, string
formulas such as "Fe2O3" are auto-converted to proper mongo queries of
{"Fe":2, "O":3}.
If 'criteria' is None, returns an empty dict. Putting this logic here
simplifies callers and allows subclasses to insert something even
when there are no criteria.
"""
if criteria is None:
return dict()
parsed_crit = dict()
for k, v in self.defaults.items():
if k not in criteria:
parsed_crit[self.aliases.get(k, k)] = v
for key, crit in list(criteria.items()):
if key in ["normalized_formula", "reduced_cell_formula"]:
comp = Composition(crit)
parsed_crit["pretty_formula"] = comp.reduced_formula
elif key == "unit_cell_formula":
comp = Composition(crit)
crit = comp.as_dict()
for el, amt in crit.items():
parsed_crit["{}.{}".format(self.aliases[key], el)] = amt
parsed_crit["nelements"] = len(crit)
parsed_crit['pretty_formula'] = comp.reduced_formula
elif key in ["$or", "$and"]:
parsed_crit[key] = [self._parse_criteria(m) for m in crit]
else:
parsed_crit[self.aliases.get(key, key)] = crit
return parsed_crit
def __init__(self, lit_file=DEFAULT_LIT_FILE):
self.line_data = {} # (line no.) -> {"formula_pretty", "tag1", "tag2", "tag3"}
if not os.path.exists(lit_file):
raise ValueError("Cannot find lit file: {}".format(lit_file))
with open(lit_file) as f:
line_no = 1
tag1 = "" # most recent level-1 tag
tag2 = "" # most recent level-2 tag
tag3 = "" # most recent level-3 tag
for line in f:
line = line.strip()
if line.startswith("###"):
tag3 = line[3:]
elif line.startswith("##"):
tag2 = line[2:]
tag3 = ""
elif line.startswith("#"):
tag1 = line[1:]
tag2 = ""
tag3 = ""
elif line:
c = Composition(line)
self.line_data[line_no] = \
{"formula_pretty": c.get_reduced_formula_and_factor()[0],
"tag1": tag1, "tag2": tag2, "tag3": tag3}
line_no += 1
def get_hhi(self, comp_or_form):
"""
Gets the reserve and production HHI for a compound.
Args:
comp_or_form (Composition or String): A Composition or String formula
Returns:
A tuple representing the (HHI_production, HHI_reserve)
"""
try:
if not isinstance(comp_or_form, Composition):
comp_or_form = Composition(comp_or_form)
hhi_p = 0
hhi_r = 0
for e in comp_or_form.elements:
percent = comp_or_form.get_wt_fraction(e)
dp, dr = self._get_hhi_el(e)
hhi_p += dp * percent
hhi_r += dr * percent
return hhi_p, hhi_r
except:
return (None, None)
def get_composition_from_string(comp_str):
"""validate and return composition from string `comp_str`."""
from pymatgen import Composition, Element
comp = Composition(comp_str)
for element in comp.elements:
Element(element)
formula = comp.get_integer_formula_and_factor()[0]
comp = Composition(formula)
return ''.join([
'{}{}'.format(key, int(value) if value > 1 else '')
for key, value in comp.as_dict().items()
])
def snl_to_wf_phonon(snl, parameters=None):
fws = []
connections = {}
parameters = parameters if parameters else {}
snl_priority = parameters.get('priority', 1)
priority = snl_priority * 2 # once we start a job, keep going!
f = Composition.from_formula(snl.structure.composition.reduced_formula).alphabetical_formula
# add the SNL to the SNL DB and figure out duplicate group
tasks = [AddSNLTask()]
spec = {'task_type': 'Add to SNL database', 'snl': snl.to_dict, '_queueadapter': QA_DB, '_priority': snl_priority}
if 'snlgroup_id' in parameters and isinstance(snl, MPStructureNL):
spec['force_mpsnl'] = snl.to_dict
spec['force_snlgroup_id'] = parameters['snlgroup_id']
del spec['snl']
fws.append(FireWork(tasks, spec, name=get_slug(f + '--' + spec['task_type']), fw_id=0))
connections[0] = [1]
# run GGA structure optimization for force convergence
spec = snl_to_wf._snl_to_spec(snl)
spec = update_spec_force_convergence(spec)
spec['run_tags'].append("origin")
spec['_priority'] = priority
spec['_queueadapter'] = QA_VASP
spec['task_type'] = "Vasp force convergence"
tasks = [VaspWriterTask(), get_custodian_task(spec)]
fws.append(FireWork(tasks, spec, name=get_slug(f + '--' + spec['task_type']), fw_id=1))
# insert into DB - GGA structure optimization
spec = {'task_type': 'VASP db insertion', '_priority': priority,
'_allow_fizzled_parents': True, '_queueadapter': QA_DB}
fws.append(
FireWork([VaspToDBTask()], spec, name=get_slug(f + '--' + spec['task_type']), fw_id=2))
connections[1] = [2]
spec = {'task_type': 'Setup Deformed Struct Task', '_priority': priority,
'_queueadapter': QA_CONTROL}
fws.append(
FireWork([SetupDeformedStructTask()], spec, name=get_slug(f + '--' + spec['task_type']),
fw_id=3))
connections[2] = [3]
wf_meta = get_meta_from_structure(snl.structure)
wf_meta['run_version'] = 'May 2013 (1)'
if '_materialsproject' in snl.data and 'submission_id' in snl.data['_materialsproject']:
wf_meta['submission_id'] = snl.data['_materialsproject']['submission_id']
return Workflow(fws, connections, name=Composition.from_formula(
snl.structure.composition.reduced_formula).alphabetical_formula, metadata=wf_meta)
def test_get_atoms(self):
if not aio.ase_loaded:
raise SkipTest("ASE not present. Skipping...")
p = Poscar.from_file(os.path.join(test_dir, 'POSCAR'))
structure = p.structure
atoms = aio.AseAtomsAdaptor.get_atoms(structure)
ase_composition = Composition.from_formula(atoms.get_name())
self.assertEqual(ase_composition, structure.composition)
def test_init(self):
filepath = os.path.join(test_dir, 'POSCAR')
poscar = Poscar.from_file(filepath)
comp = poscar.structure.composition
self.assertEqual(comp, Composition.from_formula("Fe4P4O16"))
#Vasp 4 type with symbols at the end.
poscar_string = """Test1
1.0
3.840198 0.000000 0.000000
1.920099 3.325710 0.000000
0.000000 -2.217138 3.135509
1 1
direct
0.000000 0.000000 0.000000 Si
0.750000 0.500000 0.750000 F"""
poscar = Poscar.from_string(poscar_string)
self.assertEqual(poscar.structure.composition, Composition.from_formula("SiF"))
#Vasp 4 tyle file with default names, i.e. no element symbol found.
poscar_string = """Test2
1.0
3.840198 0.000000 0.000000
1.920099 3.325710 0.000000
0.000000 -2.217138 3.135509
1 1
direct
0.000000 0.000000 0.000000
0.750000 0.500000 0.750000"""
poscar = Poscar.from_string(poscar_string)
self.assertEqual(poscar.structure.composition, Composition.from_formula("HHe"))
#Vasp 4 tyle file with default names, i.e. no element symbol found.
poscar_string = """Test3
1.0
3.840198 0.000000 0.000000
1.920099 3.325710 0.000000
0.000000 -2.217138 3.135509
1 1
Selective dynamics
direct
0.000000 0.000000 0.000000 T T T Si
0.750000 0.500000 0.750000 F F F O"""
poscar = Poscar.from_string(poscar_string)
self.assertEqual(poscar.selective_dynamics, [[True, True, True], [False, False, False]])
self.selective_poscar = poscar
def snl_to_wf(snl, do_bandstructure=True):
# TODO: clean this up once we're out of testing mode
# TODO: add WF metadata
fws = []
connections = {}
# add the SNL to the SNL DB and figure out duplicate group
tasks = [AddSNLTask()]
spec = {'task_type': 'Add to SNL database', 'snl': snl.to_dict}
fws.append(FireWork(tasks, spec, name=spec['task_type'], fw_id=0))
connections[0] = 1
# run GGA structure optimization
spec = _snl_to_spec(snl, enforce_gga=True)
tasks = [VaspWriterTask(), _get_custodian_task(spec)]
fws.append(FireWork(tasks, spec, name=spec['task_type'], fw_id=1))
# insert into DB - GGA structure optimization
spec = {'task_type': 'VASP db insertion', '_priority': 2,
'_allow_fizzled_parents': True}
spec.update(_get_metadata(snl))
fws.append(FireWork([VaspToDBTask()], spec, name=spec['task_type'], fw_id=2))
connections[1] = 2
if do_bandstructure:
spec = {'task_type': 'Controller: add Electronic Structure'}
spec.update(_get_metadata(snl))
fws.append(
FireWork([AddEStructureTask()], spec, name=spec['task_type'], fw_id=3))
connections[2] = 3
# determine if GGA+U FW is needed
incar = MPVaspInputSet().get_incar(snl.structure).to_dict
if 'LDAU' in incar and incar['LDAU']:
spec = {'task_type': 'GGA+U optimize structure (2x)',
'_dupefinder': DupeFinderVasp().to_dict()}
spec.update(_get_metadata(snl))
fws.append(FireWork(
[VaspCopyTask({'extension': '.relax2'}), SetupGGAUTask(),
_get_custodian_task(spec)], spec, name=spec['task_type'], fw_id=10))
connections[2].append(10)
spec = {'task_type': 'VASP db insertion',
'_allow_fizzled_parents': True}
spec.update(_get_metadata(snl))
fws.append(
FireWork([VaspToDBTask()], spec, name=spec['task_type'], fw_id=11))
connections[10] = 11
if do_bandstructure:
spec = {'task_type': 'Controller: add Electronic Structure'}
spec.update(_get_metadata(snl))
fws.append(FireWork([AddEStructureTask()], spec, name=spec['task_type'], fw_id=12))
connections[11] = 12
return Workflow(fws, connections, name=Composition.from_formula(snl.structure.composition.reduced_formula).alphabetical_formula)
def test_calculate_energy(self):
reactants = [Composition("MgO"), Composition("Al2O3")]
products = [Composition("MgAl2O4")]
energies = {
Composition("MgO"): -0.1,
Composition("Al2O3"): -0.2,
Composition("MgAl2O4"): -0.5
}
rxn = Reaction(reactants, products)
self.assertEqual(str(rxn), "1.000 MgO + 1.000 Al2O3 -> 1.000 MgAl2O4")
self.assertEqual(rxn.normalized_repr, "MgO + Al2O3 -> MgAl2O4")
self.assertAlmostEquals(rxn.calculate_energy(energies), -0.2, 5)
def test_remove_spectator_species(self):
rxn = BalancedReaction(
{
Composition("Li"): 4,
Composition("O2"): 1,
Composition('Na'): 1
}, {
Composition("Li2O"): 2,
Composition('Na'): 1
})
self.assertTrue(Composition('Na') not in rxn.all_comp)
def task_dict_to_wf(task_dict, launchpad):
fw_id = launchpad.get_new_fw_id()
l_id = launchpad.get_new_launch_id()
spec = {'task_type': task_dict['task_type'], 'run_tags': task_dict['run_tags'],
'vaspinputset_name': None, 'vasp': None, 'mpsnl': task_dict['snl'],
'snlgroup_id': task_dict['snlgroup_id']}
tasks = [DummyLegacyTask()]
launch_dir = task_dict['dir_name_full']
stored_data = {'error_list': []}
update_spec = {'prev_vasp_dir': task_dict['dir_name'],
'prev_task_type': spec['task_type'],
'mpsnl': spec['mpsnl'], 'snlgroup_id': spec['snlgroup_id'],
'run_tags': spec['run_tags']}
fwaction = FWAction(stored_data=stored_data, update_spec=update_spec)
if task_dict['completed_at']:
complete_date = datetime.datetime.strptime(task_dict['completed_at'], "%Y-%m-%d %H:%M:%S")
state_history = [{"created_on": complete_date, 'state': 'COMPLETED'}]
else:
state_history = []
launches = [Launch('COMPLETED', launch_dir, fworker=None, host=None, ip=None, action=fwaction,
state_history=state_history, launch_id=l_id, fw_id=fw_id)]
f = Composition.from_formula(task_dict['pretty_formula']).alphabetical_formula
fw = FireWork(tasks, spec, name=get_slug(f + '--' + spec['task_type']), launches=launches, state='COMPLETED', created_on=None,
fw_id=fw_id)
wf_meta = get_meta_from_structure(Structure.from_dict(task_dict['snl']))
wf_meta['run_version'] = 'preproduction (0)'
wf = Workflow.from_FireWork(fw, name=f, metadata=wf_meta)
launchpad.add_wf(wf, reassign_all=False)
launchpad._upsert_launch(launches[0])
print 'ADDED', fw_id
# return fw_id
return fw_id
def synth_score_from_formula(self, formula):
"""
Takes input in form of an unreduced crystal composition and returns
the synthesizability score for the same.
Args:
formula (str): For example: 'Ba2Yb2Al4Si2N10O4'
Returns:
synth_scores (list): Synthesizability scores (between 0 and 1) of
given sample.
"""
# Updating the criteria
updated_input = {"full_formula": formula}
# Extracting the composition
comp = Composition(formula)
# Extracting the MP-IDs
mp_id = self._get_mp_id(updated_input)
# If Input given is invalid
if len(mp_id) == 0:
return "No such compound exists in Materials Project Database"
# Checking if it already exists in pre-trained model
fblock, mp_id, synth_score, df_input = self._check_if_already_exists(
comp, updated_input)
# If found in pre-trained models, report synth_scores
if df_input.shape[0] == 0:
if len(synth_score) != 0:
logging.info("The synthesizability scores are %s", synth_score)
return synth_score
# Else, train the model and report scores
else:
df_extracted = self._extract_features(df_input)
input_ids, df = self._append_data(df_extracted, fblock)
score = self._do_pulearner(input_ids, df)
synth_score.append(score)
logging.info("The synthesizability scores are %s", synth_score)
return synth_score
def testName(self):
entry_Li = ComputedEntry("Li", -1.90753119)
with open(os.path.join(test_dir, "LiTiO2_batt.json"), "r") as f:
entries_LTO = json.load(f, cls=MontyDecoder)
ie_LTO = InsertionElectrode.from_entries(entries_LTO, entry_Li)
with open(os.path.join(test_dir, "FeF3_batt.json"), "r") as fid:
entries = json.load(fid, cls=MontyDecoder)
ce_FF = ConversionElectrode.from_composition_and_entries(
Composition("FeF3"), entries
)
plotter = VoltageProfilePlotter(xaxis="frac_x")
plotter.add_electrode(ie_LTO, "LTO insertion")
plotter.add_electrode(ce_FF, "FeF3 conversion")
self.assertIsNotNone(plotter.get_plot_data(ie_LTO))
self.assertIsNotNone(plotter.get_plot_data(ce_FF))
def get_polymorphs(f_str, pda):
"""get_polymorphs
Gets all the polymorphs with a chemical composition within the cutoff.
:param f_str: Chemical composition we're interested in.
:param pda: A PDAnalyzer for the appropriate phase diagram.
:return: A list of entries within the cutoff.
"""
comp = Composition(f_str).reduced_composition
relevant_polymorphs = []
for e in pda._pd.all_entries:
if e.composition.reduced_composition == comp:
ehull = pda.get_decomp_and_e_above_hull(e)[1]
anion = [el for el in e.composition.as_dict() if el in anions][0]
if ehull < ehull_cut[anion]:
form_e = pda._pd.get_form_energy_per_atom(e)
e.data["formation_energy_per_atom"] = form_e
relevant_polymorphs.append(e)
return relevant_polymorphs
def scrape_comp(row):
# reformulate the doped site by electronegativity order of elements for a better web scraping,and remove '0's & parantheses. e.g. Pb(Ti0.50Zr0.50)O3 --> PbZr0.5Ti0.5O3
site_list = re.split(
r'[()]', row.StructuredFormula
) # get what's inside parantheses and others separately
doped_site = max(
site_list,
key=len) # get doped site re.sub(r"(?<=\d)0+", "", d)
site_rearranged = '(' + re.sub(
r"(?<=\d)0+", "",
Composition(doped_site).formula.replace(' ', '')
) + ')' # arrange elements in electronegative order, remove '0' and club with ()
site_list[site_list.index(
doped_site)] = site_rearranged # replace with processed site
compound = str(''.join(site_list))
# print (compound)
result = self.search(compound)
return compound, result
def get_meta_from_structure(structure):
if isinstance(structure, TransformedStructure):
structure = structure.final_structure
comp = structure.composition
elsyms = sorted(set([e.symbol for e in comp.elements]))
meta = {'nsites': structure.num_sites,
'elements': elsyms,
'nelements': len(elsyms),
'formula': comp.formula,
'formula_pretty': comp.reduced_formula,
'formula_reduced_abc': Composition(comp.reduced_formula)
.alphabetical_formula,
'formula_anonymous': comp.anonymized_formula,
'chemsys': '-'.join(elsyms),
'is_ordered': structure.is_ordered,
'is_valid': structure.is_valid()}
return meta
def __init__(self, mpid, cmpd_dict):
"""Initializing the feature finder.
Sets up a FeatureFinder based on a dict from the hydrate finder.
Args:
mpid (str): The MP ID of the dry compound in question.
cmpd_dict (dict): A dictionary output from the hydrate finder.
"""
comp = Composition(cmpd_dict['formula']).as_dict()
self.comp = comp
self.num_atoms = sum([comp[ele] for ele in comp])
self.mpid = mpid
with MPRester() as mpr:
struct = mpr.get_structure_by_material_id(self.mpid)
self.struct = struct
def get_composition_from_string(comp_str):
"""validate and return composition from string `comp_str`."""
from pymatgen import Composition, Element
comp = Composition(comp_str)
for element in comp.elements:
Element(element)
formula = comp.get_integer_formula_and_factor()[0]
comp = Composition(formula)
return ''.join([
'{}{}'.format(key,
int(value) if value > 1 else '')
for key, value in comp.as_dict().items()
])
def convert_mpworks_to_atomate(mpworks_doc, update_mpworks=True):
"""
Function to convert an mpworks document into an atomate
document, uses schema above and a few custom cases
Args:
mpworks_doc (dict): mpworks task document
update_mpworks (bool): flag to indicate that mpworks schema
should be updated to final MPWorks version
"""
if update_mpworks:
update_mpworks_schema(mpworks_doc)
atomate_doc = {}
for key_mpworks, key_atomate in settings['task_conversion_keys'].items():
val = get(mpworks_doc, key_mpworks)
set_(atomate_doc, key_atomate, val)
# Task type
atomate_doc["task_label"] = settings['task_label_conversions'].get(
mpworks_doc["task_type"])
# calculations
atomate_doc["calcs_reversed"] = mpworks_doc["calculations"][::-1]
# anonymous formula
comp = Composition(atomate_doc['composition_reduced'])
atomate_doc["formula_anonymous"] = comp.anonymized_formula
# deformation matrix and original_task_id
if "deformation_matrix" in mpworks_doc:
# Transpose this b/c of old bug, should verify in doc processing
defo = mpworks_doc["deformation_matrix"]
if isinstance(defo, str):
defo = convert_string_deformation_to_list(defo)
defo = np.transpose(defo).tolist()
set_(atomate_doc, "transmuter.transformations",
["DeformStructureTransformation"])
set_(atomate_doc, "transmuter.transformation_params",
[{
"deformation": defo
}])
return atomate_doc
def test_hse_structure_opt(default_dict):
default_dict.update({
"xc": Xc.hse,
"composition": Composition("UO2"),
"symbol_list": ["U", "O"],
"potcar": Potcar(["U", "O"]),
"exchange_ratio": 0.5,
"set_hubbard_u": True
})
generator = IncarSettingsGenerator(**default_dict)
# for k, v in generator.incar_settings.items():
# print(f'\"{k}\": {v}, ')
expected = {
"ALGO": "Damped",
"PREC": "Normal",
"LREAL": False,
"EDIFF": 1e-07,
"ENCUT": 520.0,
"LASPH": True,
"NELM": 100,
"ISIF": 3,
"IBRION": 2,
"EDIFFG": -0.005,
"NSW": 50,
"ISMEAR": 0,
"SIGMA": 0.1,
"LWAVE": True,
"LCHARG": False,
"LORBIT": 10,
"LHFCALC": True,
"PRECFOCK": "Fast",
"TIME": 0.4,
"AEXX": 0.5,
"HFSCREEN": 0.208,
"LDAU": True,
"LDAUTYPE": 2,
"LMAXMIX": 6,
"LDAUPRINT": 1,
"LDAUU": [5, 0],
"LDAUL": [3, -1],
"KPAR": 1,
}
assert generator.incar_settings == expected
def setUp(self):
self.formulas = ["LiCoO2", "FeF3", "MnO2"]
self.conversion_eletrodes = {}
for f in self.formulas:
with open(
os.path.join(PymatgenTest.TEST_FILES_DIR,
f + "_batt.json"), "r") as fid:
entries = json.load(fid, cls=MontyDecoder)
if f in ["LiCoO2", "FeF3"]:
working_ion = "Li"
elif f in ["MnO2"]:
working_ion = "Mg"
c = ConversionElectrode.from_composition_and_entries(
Composition(f), entries, working_ion_symbol=working_ion)
self.conversion_eletrodes[f] = {
"working_ion": working_ion,
"CE": c
}
self.expected_properties = {
"LiCoO2": {
"average_voltage": 2.26940307125,
"capacity_grav": 903.19752911225669,
"capacity_vol": 2903.35804724,
"specific_energy": 2049.7192465127678,
"energy_density": 6588.8896693479574,
},
"FeF3": {
"average_voltage": 3.06179925889,
"capacity_grav": 601.54508701578118,
"capacity_vol": 2132.2069115142394,
"specific_energy": 1841.8103016131706,
"energy_density": 6528.38954147,
},
"MnO2": {
"average_voltage": 1.7127027687901726,
"capacity_grav": 790.9142070034802,
"capacity_vol": 3543.202003526853,
"specific_energy": 1354.6009522103434,
"energy_density": 6068.451881823329,
},
}
def screen_formulas(candidates: Iterable[str]) -> Iterable[str]:
"""[Filters unique chemical compositions]
Arguments:
candidates {Iterable[str]} -- [List of chemical formula candidates]
Returns:
Iterable[str] -- [Filtered list of candidates]
"""
try:
formulas = []
for formula in candidates:
comp = Composition(formula)
if comp.reduced_formula not in formulas:
formulas.append(comp.reduced_formula)
return formulas
except Exception as exc:
print('-- Could not filter candidates --')
raise (exc)
def get_magpie_descriptor(comp, descriptor_name):
"""
Get descriptor data for elements in a compound from the Magpie data repository.
Args:
comp: (str) compound composition, eg: "NaCl"
descriptor_name: name of Magpie descriptor needed. Find the entire list at
https://bitbucket.org/wolverton/magpie/src/6ecf8d3b79e03e06ef55c141c350a08fbc8da849/Lookup%20Data/?at=master
Returns: (list) of descriptor values for each element in the composition
"""
magpiedata_lst = []
magpiedata = collections.namedtuple(
'magpiedata', 'element propname propvalue propunit amt')
available_props = []
for datafile in os.listdir('data/magpie_elementdata'):
available_props.append(datafile.replace('.table', ''))
if descriptor_name not in available_props:
raise ValueError(
"This descriptor is not available from the Magpie repository. Choose from {}"
.format(available_props))
el_amt = Composition(comp).get_el_amt_dict()
unit = None
with open('data/magpie_elementdata/README.txt', 'r') as readme_file:
readme_file_line = readme_file.readlines()
for lineno, line in enumerate(readme_file_line, 1):
if descriptor_name + '.table' in line:
if 'Units: ' in readme_file_line[lineno + 1]:
unit = readme_file_line[lineno +
1].split(':')[1].strip('\n')
with open('data/magpie_elementdata/' + descriptor_name + '.table',
'r') as descp_file:
lines = descp_file.readlines()
for el in el_amt:
atomic_no = Element(el).Z
magpiedata_lst.append(
magpiedata(element=el,
propname=descriptor_name,
propvalue=float(lines[atomic_no - 1]),
propunit=unit,
amt=el_amt[el]))
return magpiedata_lst
def featurize(self,
compositions: Iterable[str],
log_every_n: int = 1000) -> np.ndarray:
"""Calculate features for crystal compositions.
Parameters
----------
compositions: Iterable[str]
Iterable sequence of composition strings, e.g. "MoS2".
log_every_n: int, default 1000
Logging messages reported every `log_every_n` samples.
Returns
-------
features: np.ndarray
A numpy array containing a featurized representation of
`compositions`.
"""
# Convert iterables to list
compositions = list(compositions)
try:
from pymatgen import Composition
except ModuleNotFoundError:
raise ValueError("This class requires pymatgen to be installed.")
features = []
for idx, composition in enumerate(compositions):
if idx % log_every_n == 0:
logger.info("Featurizing datapoint %i" % idx)
try:
c = Composition(composition)
features.append(self._featurize(c))
except:
logger.warning(
"Failed to featurize datapoint %i. Appending empty array" %
idx)
features.append(np.array([]))
features = np.asarray(features)
return features
def test_column_attr(self):
"""
Test that the autofeaturizer object correctly takes in composition_col,
structure_col, bandstruct_col, and dos_col, and checks that
fit_and_transform()
works correctly with the attributes.
"""
# Modification of test_featurize_composition with AutoFeaturizer parameter
target = "K_VRH"
df = copy.copy(self.test_df[['composition', target]].iloc[:self.limit])
af = AutoFeaturizer(composition_col="composition", preset="best", ignore_errors=False)
df = af.fit_transform(df, target)
self.assertEqual(df["LUMO_element"].iloc[0], "Nb")
self.assertTrue("composition" not in df.columns)
df = self.test_df[["composition", target]].iloc[:self.limit]
df["composition"] = [Composition(s) for s in df["composition"]]
af = AutoFeaturizer(composition_col="composition", preset="best")
df = af.fit_transform(df, target)
self.assertEqual(df["LUMO_element"].iloc[0], "Nb")
self.assertTrue("composition" not in df.columns)
# Modification of test_featurize_structure with AutoFeaturizer parameter
target = "K_VRH"
df = copy.copy(self.test_df[['structure', target]].iloc[:self.limit])
af = AutoFeaturizer(structure_col="structure", preset="fast")
df = af.fit_transform(df, target)
self.assertTrue("vpa" in df.columns)
self.assertTrue("HOMO_character" in df.columns)
self.assertTrue("composition" not in df.columns)
self.assertTrue("structure" not in df.columns)
df = copy.copy(self.test_df[['structure', target]].iloc[:self.limit])
df["structure"] = [s.as_dict() for s in df["structure"]]
af = AutoFeaturizer(structure_col="structure", preset="fast")
df = af.fit_transform(df, target)
self.assertTrue("vpa" in df.columns)
self.assertTrue("HOMO_character" in df.columns)
self.assertTrue("composition" not in df.columns)
self.assertTrue("structure" not in df.columns)
def test_from_string(self):
rxn = BalancedReaction({
Composition("Li"): 4,
Composition("O2"): 1
}, {Composition("Li2O"): 2})
self.assertEqual(rxn,
BalancedReaction.from_string("4 Li + O2 -> 2Li2O"))
rxn = BalancedReaction({Composition("Li(NiO2)3"): 1}, {
Composition("O2"): 0.5,
Composition("Li(NiO2)2"): 1,
Composition("NiO"): 1
})
self.assertEqual(
rxn,
BalancedReaction.from_string(
"1.000 Li(NiO2)3 -> 0.500 O2 + 1.000 Li(NiO2)2 + 1.000 NiO"))
def get_entries(self, chemsys):
"""
Get all entries in a chemsys from materials
Args:
chemsys(str): a chemical system represented by string elements seperated by a dash (-)
Returns:
set(ComputedEntry): a set of entries for this system
"""
self.logger.info("Getting entries for: {}".format(chemsys))
new_q = dict(self.query)
new_q["chemsys"] = {"$in": list(chemsys_permutations(chemsys))}
fields = [
"structure", self.materials.key, "thermo.energy_per_atom",
"composition", "calc_settings"
]
data = list(self.materials.query(fields, new_q))
all_entries = []
for d in data:
comp = Composition(d["composition"])
entry = ComputedEntry(
comp,
d["thermo"]["energy_per_atom"] * comp.num_atoms,
0.0,
parameters=d["calc_settings"],
entry_id=d[self.materials.key],
data={
"oxide_type":
oxide_type(Structure.from_dict(d["structure"]))
})
all_entries.append(entry)
self.logger.info("Total entries in {} : {}".format(
chemsys, len(all_entries)))
return all_entries
def get_chemsys(formula_or_structure, seperator='-'):
"""
Gets a sorted, character-delimited set of elements, e.g.
Fe-Ni-O or O-Ti
Args:
formula_or_structure (str, Structure): formula or structure
for which to get chemical system
separator (str): separator for the chemsys elements
Returns:
(str): separated
"""
if isinstance(formula_or_structure, Structure):
formula = formula_or_structure.composition.reduced_formula
else:
formula = formula_or_structure
elements = [str(el) for el in Composition(formula)]
return seperator.join(sorted(elements))
def get_atomic_lists_and_numbers(compositions):
# get atom number
atom_lists = np.empty(118, dtype=object)
atoms_duplicated = []
atomic_numbers_duplicated = []
for line in compositions:
if line == 'nan':
line = 'NaN'
atom_list = Composition(line[0])
#atom_list = Composition(line)
for element in atom_list:
atomic_numbers_duplicated.append(element.Z - 1)
atom_lists[element.Z - 1] = element.symbol
# reduce duplicate
atomic_numbers = np.array(list(set(atomic_numbers_duplicated)))
atomic_numbers = np.sort(atomic_numbers, axis=0)
return atom_lists, atomic_numbers
def test_prediction(self):
sp = SubstitutionPredictor(threshold = 8e-3)
result = sp.list_prediction(['Na+', 'Cl-'], to_this_composition = True)[5]
cprob = sp.p.cond_prob_list(result['substitutions'].keys(),
result['substitutions'].values())
self.assertAlmostEqual(result['probability'], cprob)
self.assertEqual(set(result['substitutions'].values()), set(['Na+', 'Cl-']))
result = sp.list_prediction(['Na+', 'Cl-'], to_this_composition = False)[5]
cprob = sp.p.cond_prob_list(result['substitutions'].keys(),
result['substitutions'].values())
self.assertAlmostEqual(result['probability'], cprob)
self.assertNotEqual(set(result['substitutions'].values()),
set(['Na+', 'Cl-']))
c = Composition({'Ag2+' : 1, 'Cl-' : 2})
result = sp.composition_prediction(c, to_this_composition = True)[2]
self.assertEqual(set(result['substitutions'].values()), set(c.elements))
result = sp.composition_prediction(c, to_this_composition = False)[2]
self.assertEqual(set(result['substitutions'].keys()), set(c.elements))
def test_featurize_composition(self):
"""
Test automatic featurization while only considering formula/composition.
"""
target = "K_VRH"
# When compositions are strings
df = copy.copy(self.test_df[["composition", target]].iloc[: self.limit])
af = AutoFeaturizer(preset="express")
df = af.fit_transform(df, target)
self.assertAlmostEqual(df["MagpieData minimum Number"].iloc[2], 14.0)
self.assertTrue("composition" not in df.columns)
# When compositions are Composition objects
df = self.test_df[["composition", target]].iloc[: self.limit]
df["composition"] = [Composition(s) for s in df["composition"]]
af = AutoFeaturizer(preset="express")
df = af.fit_transform(df, target)
self.assertAlmostEqual(df["MagpieData minimum Number"].iloc[2], 14.0)
self.assertTrue("composition" not in df.columns)
def test_energy_window(mocker):
mock_bs = mocker.MagicMock()
mock_bs.efermi = 0
mock_bs.is_metal.return_value = True
stub_vasprun = mocker.MagicMock()
stub_vasprun.final_structure.composition = Composition("MgO2")
stub_vasprun.get_band_structure.return_value = mock_bs
energy = {
"1": [
np.array([[-0.2, -0.1, -0.3, -0.1], [-0.2, -0.1, -0.3, 0.1],
[1.2, 1.3, 1.1, 1.2]])
]
}
distances = [[0.0, 1.0]]
labels = ["A", "GAMMA"]
label_distances = [0.0, 1.0]
plot_data = {
"energy": energy,
"distances": distances,
"vbm": None,
"cbm": None
}
mock_bsp = mocker.patch("vise.analyzer.vasp.plot_band.BSPlotter",
auto_spec=True)
mock_bsp.return_value.bs_plot_data.return_value = plot_data
mock_bsp.return_value.get_ticks_old.return_value = {
"label": labels,
"distance": distances[0]
}
plot_info = BandPlotInfoFromVasp(stub_vasprun,
"KPOINTS",
energy_window=[0.0, 1.0
]).make_band_plot_info()
assert (plot_info.band_info_set[0].band_energies == [[[[
-0.2, -0.1, -0.3, 0.1
]]]])
def compute_grav_cap(comp, wi, remove=True):
"""compute_grav_cap
Computes the gravimetric capacity of a composition.
:param comp: Composition dictionary object.
:param wi: String which is the symbol of the working ion.
:param remove: Are we removing? If not, we're inserting.
:return: Gravimetric capacity in mAh/g.
"""
c = comp.as_dict()
weight = comp.weight
wi_charge = mobile_ion_charges[wi]
if remove:
num_wi = max_removal(c)
else:
num_wi = max_insertion(c, wi)
wi_comp = Composition(wi)
weight += wi_comp.weight * num_wi
print num_wi, wi_charge
cap = num_wi * wi_charge * ELECTRON_TO_AMPERE_HOURS
grav_cap = cap / (weight / 1000)
return grav_cap
def get_meta_from_structure(structure):
# TODO: this won't work for molecules
meta = {}
meta['nsites'] = len(structure.sites)
meta['elements'] = list(set([el.symbol for el in structure.composition.elements]))
meta['nelements'] = len(meta['elements'])
meta['formula'] = structure.composition.formula
meta['formula_red'] = structure.composition.reduced_formula
meta['formula_abc_red'] = Composition.from_formula(structure.composition.reduced_formula).alphabetical_formula
meta['composition_dict'] = structure.composition.to_dict
meta['anonymized_formula'] = structure.composition.anonymized_formula
meta['chemsystem'] = '-'.join(sorted(list(set([e.symbol for e in structure.composition.elements])))) # the complex logic set/list is to prevent duplicates if there are multiple oxidation states
meta['is_ordered'] = structure.is_ordered
#promixity warning:
meta['proximity_warning'] = False
for (s1, s2) in itertools.combinations(structure._sites, 2):
if s1.distance(s2) < Structure.DISTANCE_TOLERANCE:
meta['proximity_warning'] = True
return meta
def test_to_from_dict(self):
rct = {
Composition('K2SO4'): 3,
Composition('Na2S'): 1,
Composition('Li'): 24
}
prod = {
Composition('KNaS'): 2,
Composition('K2S'): 2,
Composition('Li2O'): 12
}
rxn = BalancedReaction(rct, prod)
d = rxn.as_dict()
new_rxn = BalancedReaction.from_dict(d)
for comp in new_rxn.all_comp:
self.assertEqual(new_rxn.get_coeff(comp), rxn.get_coeff(comp))
def create_df():
newcoll = db['abc3']
newcoll.drop()
x = 0
for doc in db['pauling_file_min_tags'].find().batch_size(75):
x += 1
if x % 1000 == 0:
print x
if doc['metadata']['_structure']['anonymized_formula'] == 'ABC3' and doc['is_ht'] in [True,False] \
and 'TiO3' in doc['metadata']['_structure']['reduced_cell_formula']:
newcoll.insert(doc)
cursor = newcoll.find()
df = pd.DataFrame(list(cursor))
for i, row in df.iterrows():
df.set_value(i, 'reduced_cell_formula',
row['metadata']['_structure']['reduced_cell_formula'])
try:
df.set_value(
i, 'space_group',
int(row['metadata']['_Springer']['geninfo']['Space Group']))
except:
df.set_value(i, 'space_group', None)
try:
df.set_value(
i, 'density',
float(row['metadata']['_Springer']['geninfo']
['Density'].split()[2]))
except IndexError as e:
df.set_value(i, 'density', None)
structure = Structure.from_dict(row['structure'])
num_density = (structure.num_sites / structure.volume)
no_of_atoms = Composition(structure.composition).num_atoms
num_vol = (structure.volume / no_of_atoms)
df.set_value(i, 'number_density', num_density)
df.set_value(i, 'number_volume', num_vol)
if row['metadata']['_structure']['is_ordered']:
df.set_value(i, 'is_ordered', 1)
else:
df.set_value(i, 'is_ordered', 0)
df.to_pickle('abc3.pkl')
def test_vasp_band_plotter(is_metal, expected_band_edge, mocker):
mock_bs = mocker.MagicMock()
mock_bs.efermi = 10
mock_bs.is_metal.return_value = is_metal
stub_vasprun = mocker.MagicMock()
stub_vasprun.final_structure.composition = Composition("MgO2")
stub_vasprun.get_band_structure.return_value = mock_bs
energy = {"1": [np.array([[0.1], [0.2], [0.3]])]}
distances = [np.array([0.0, 0.1, 0.2])]
labels = ["A", "$A_0$", "GAMMA"]
label_distances = [0.0, 0.1, 0.2]
plot_data = {
"energy": energy,
"distances": distances,
"vbm": [[0, -100]],
"cbm": [[1, 100], [2, 100]]
}
mock_bsp = mocker.patch("vise.analyzer.vasp.plot_band.BSPlotter",
auto_spec=True)
mock_bsp.return_value.bs_plot_data.return_value = plot_data
mock_bsp.return_value.get_ticks_old.return_value = {
"label": labels,
"distance": [0.0, 0.1, 0.2]
}
plot_info = BandPlotInfoFromVasp(stub_vasprun,
"KPOINTS").make_band_plot_info()
expected_x_ticks = XTicks(labels=["A", "${\\rm A}_0$", "Γ"],
distances=label_distances)
assert plot_info.band_info_set[0].band_energies == [[[[0.1], [0.2],
[0.3]]]]
assert plot_info.band_info_set[0].band_edge == expected_band_edge
assert plot_info.distances_by_branch == [[0.0, 0.1, 0.2]]
assert plot_info.x_ticks == expected_x_ticks
assert plot_info.title == "MgO$_{2}$"
def get_random_structure():
print("input the formula of structure like this: ")
print("(Fe3O4)4")
print("or input the formula and spacegroup like this ")
print("Si4O8 20")
wait_sep()
in_str=""
while in_str=="":
in_str=input().strip().split()
if len(in_str)==1:
spgRange=[1,230]
tag=False
elif len(in_str)==2:
spgRange=[int(in_str[1]),int(in_str[1])+1]
tag=True
else:
print("unknow format")
os._exit()
comp=Composition(in_str[0])
elem=[el.symbol for el in comp]
num_atom=[int(comp[el]) for el in elem]
i=124456
scale=1.5
while True:
random.seed(i)
spg=np.random.randint(spgRange[0],spgRange[1])
print("try space group %s"%(spg))
i+=1
rc=random_crystal(spg,elem, num_atom, scale)
pmg_structure=rc.prim_struct
if pmg_structure is not None:
print("Generate random structure with spg: %d natom: %d"%(spg, len(pmg_structure)))
print(pmg_structure)
pmg_structure.to('poscar','random_spg_'+str(spg)+'.vasp')
return
if tag:
print("Failed to generate structure with spg: %d"%(spg))
return
def get_meta_from_structure(structure):
"""
Used by `structure_to_mock_job`, to "fill out" a job document.
:param structure: pymatgen structure object
:return: (dict) structure metadata
"""
comp = structure.composition
elsyms = sorted(set([e.symbol for e in comp.elements]))
meta = {
'nsites': len(structure),
'elements': elsyms,
'nelements': len(elsyms),
'formula': comp.formula,
'reduced_cell_formula': comp.reduced_formula,
'reduced_cell_formula_abc':
Composition(comp.reduced_formula).alphabetical_formula,
'anonymized_formula': comp.anonymized_formula,
'chemsystem': '-'.join(elsyms),
'is_ordered': structure.is_ordered,
'is_valid': structure.is_valid(tol=0.50)
}
return meta
def snl_to_wf_ggau(snl):
# TODO: add WF meta
fws = []
connections = {}
# add the root FW (GGA+U)
spec = _snl_to_spec(snl, enforce_gga=False)
tasks = [VaspWriterTask(), _get_custodian_task(spec)]
fws.append(FireWork(tasks, spec, fw_id=1))
# add GGA insertion to DB
spec = {'task_type': 'VASP db insertion', '_priority': 2,
'_category': 'VASP'}
spec.update(_get_metadata(snl))
fws.append(FireWork([VaspToDBTask()], spec, fw_id=2))
connections[1] = 2
mpvis = MPVaspInputSet()
spec['vaspinputset_name'] = mpvis.__class__.__name__
return Workflow(fws, connections, name=Composition.from_formula(snl.structure.composition.reduced_formula).alphabetical_formula)
def test_init(self):
filepath = os.path.join(test_dir, 'input/23221-ZDsSsJoEW14.res')
res = Res.from_file(filepath)
comp = res.structure.composition
self.assertEqual(comp, Composition.from_formula("C194H60"))
#print res
res_string = """TITL
CELL 1.0 1.0 1.0 1.0 90.0 90.0 90.0
LATT -1
SFAC Si F
Si 1 0.000000 0.000000 0.000000 1.0
F 2 0.750000 0.500000 0.750000 1.0"""
res = Res.from_string(res_string)
self.assertEqual(res.structure.composition, Composition("SiF"))
self.assertEquals(res.structure.num_sites, 2)
#print res
struct = Structure(Lattice.orthorhombic(2.5, 3.5, 7.0), ['Na', 'Cl'], [[1.0, 0.0, 0.0], [0.0, 1.0, 0.0]])
res = Res(struct)
res_string = str(res)
lines = res_string.splitlines()
self.assertEqual(lines[1], "CELL 1.0 2.5 3.5 7.0 90.0 90.0 90.0")
def test_to_from_dict(self):
d = self.vinput.to_dict
vinput = VaspInput.from_dict(d)
comp = vinput["POSCAR"].structure.composition
self.assertEqual(comp, Composition.from_formula("Fe4P4O16"))
def snl_to_wf(snl, parameters=None):
fws = []
connections = {}
parameters = parameters if parameters else {}
snl_priority = parameters.get('priority', 1)
priority = snl_priority * 2 # once we start a job, keep going!
f = Composition.from_formula(snl.structure.composition.reduced_formula).alphabetical_formula
# add the SNL to the SNL DB and figure out duplicate group
tasks = [AddSNLTask()]
spec = {'task_type': 'Add to SNL database', 'snl': snl.to_dict, '_queueadapter': QA_DB, '_priority': snl_priority}
if 'snlgroup_id' in parameters and isinstance(snl, MPStructureNL):
spec['force_mpsnl'] = snl.to_dict
spec['force_snlgroup_id'] = parameters['snlgroup_id']
del spec['snl']
fws.append(FireWork(tasks, spec, name=get_slug(f + '--' + spec['task_type']), fw_id=0))
connections[0] = [1]
# run GGA structure optimization
spec = _snl_to_spec(snl, enforce_gga=True)
spec['_priority'] = priority
spec['_queueadapter'] = QA_VASP
tasks = [VaspWriterTask(), get_custodian_task(spec)]
fws.append(FireWork(tasks, spec, name=get_slug(f + '--' + spec['task_type']), fw_id=1))
# insert into DB - GGA structure optimization
spec = {'task_type': 'VASP db insertion', '_priority': priority,
'_allow_fizzled_parents': True, '_queueadapter': QA_DB}
fws.append(
FireWork([VaspToDBTask()], spec, name=get_slug(f + '--' + spec['task_type']), fw_id=2))
connections[1] = [2]
if not parameters.get('skip_bandstructure', False):
spec = {'task_type': 'Controller: add Electronic Structure v2', '_priority': priority,
'_queueadapter': QA_CONTROL}
fws.append(
FireWork([AddEStructureTask()], spec, name=get_slug(f + '--' + spec['task_type']),
fw_id=3))
connections[2] = [3]
# determine if GGA+U FW is needed
incar = MPVaspInputSet().get_incar(snl.structure).to_dict
if 'LDAU' in incar and incar['LDAU']:
spec = _snl_to_spec(snl, enforce_gga=False)
del spec['vasp'] # we are stealing all VASP params and such from previous run
spec['_priority'] = priority
spec['_queueadapter'] = QA_VASP
fws.append(FireWork(
[VaspCopyTask(), SetupGGAUTask(),
get_custodian_task(spec)], spec, name=get_slug(f + '--' + spec['task_type']),
fw_id=10))
connections[2].append(10)
spec = {'task_type': 'VASP db insertion', '_queueadapter': QA_DB,
'_allow_fizzled_parents': True, '_priority': priority}
fws.append(
FireWork([VaspToDBTask()], spec, name=get_slug(f + '--' + spec['task_type']), fw_id=11))
connections[10] = [11]
if not parameters.get('skip_bandstructure', False):
spec = {'task_type': 'Controller: add Electronic Structure v2', '_priority': priority,
'_queueadapter': QA_CONTROL}
fws.append(
FireWork([AddEStructureTask()], spec, name=get_slug(f + '--' + spec['task_type']),
fw_id=12))
connections[11] = [12]
wf_meta = get_meta_from_structure(snl.structure)
wf_meta['run_version'] = 'May 2013 (1)'
if '_materialsproject' in snl.data and 'submission_id' in snl.data['_materialsproject']:
wf_meta['submission_id'] = snl.data['_materialsproject']['submission_id']
return Workflow(fws, connections, name=Composition.from_formula(
snl.structure.composition.reduced_formula).alphabetical_formula, metadata=wf_meta)
if x % 1000 == 0:
print x
if (doc['metadata']['_Springer']['geninfo']['Refined Formula']).replace(u'\u2013', '-') != '-':
form = doc['metadata']['_Springer']['geninfo']['Refined Formula']
else:
form = doc['metadata']['_Springer']['geninfo']['Alphabetic Formula']
form = form.replace('[', '')
form = form.replace(']', '')
try:
redform = Composition(form).reduced_formula
except:
print 'Could not parse composition for key:{} with composition:{}'.format(doc['key'], form)
y += 1
continue
alphaform_comp = Composition(redform).alphabetical_formula
alphaform_frac = Composition(alphaform_comp).fractional_composition
structure_comp = doc['metadata']['_structure']['reduced_cell_formula_abc']
structure_frac = Composition(structure_comp).fractional_composition
if alphaform_frac != structure_frac and not alphaform_frac.almost_equals(structure_frac, 0.15):
for document in db['pauling_file'].find({'key': doc['key']}):
try:
struct_lst = CifParser.from_string(document['cif_string']).get_structures()
except:
pass
if len(struct_lst) > 1:
matched = False
for struct in struct_lst[1:]:
another_frac = Composition(struct.composition).fractional_composition
if alphaform_frac == another_frac or alphaform_frac.almost_equals(another_frac, 0.15):
matched = True
# '''
def run_task(self, fw_spec):
print 'sleeping 10s for Mongo'
time.sleep(10)
print 'done sleeping'
print 'the gap is {}, the cutoff is {}'.format(fw_spec['analysis']['bandgap'], self.gap_cutoff)
if fw_spec['analysis']['bandgap'] >= self.gap_cutoff:
print 'Adding more runs...'
type_name = 'GGA+U' if 'GGA+U' in fw_spec['prev_task_type'] else 'GGA'
snl = StructureNL.from_dict(fw_spec['mpsnl'])
f = Composition.from_formula(snl.structure.composition.reduced_formula).alphabetical_formula
fws = []
connections = {}
priority = fw_spec['_priority']
# run GGA static
spec = fw_spec # pass all the items from the current spec to the new
# one
spec.update({'task_type': '{} static'.format(type_name), '_queueadapter': QA_VASP,
'_dupefinder': DupeFinderVasp().to_dict(), '_priority': priority})
fws.append(
FireWork(
[VaspCopyTask({'use_CONTCAR': True}), SetupStaticRunTask(),
get_custodian_task(spec)], spec, name=get_slug(f+'--'+spec['task_type']), fw_id=-10))
# insert into DB - GGA static
spec = {'task_type': 'VASP db insertion', '_queueadapter': QA_DB,
'_allow_fizzled_parents': True, '_priority': priority}
fws.append(
FireWork([VaspToDBTask()], spec, name=get_slug(f+'--'+spec['task_type']), fw_id=-9))
connections[-10] = -9
# run GGA Uniform
spec = {'task_type': '{} Uniform'.format(type_name), '_queueadapter': QA_VASP,
'_dupefinder': DupeFinderVasp().to_dict(), '_priority': priority}
fws.append(FireWork(
[VaspCopyTask({'use_CONTCAR': False}), SetupNonSCFTask({'mode': 'uniform'}),
get_custodian_task(spec)], spec, name=get_slug(f+'--'+spec['task_type']), fw_id=-8))
connections[-9] = -8
# insert into DB - GGA Uniform
spec = {'task_type': 'VASP db insertion', '_queueadapter': QA_DB,
'_allow_fizzled_parents': True, '_priority': priority}
fws.append(
FireWork([VaspToDBTask({'parse_uniform': True})], spec, name=get_slug(f+'--'+spec['task_type']),
fw_id=-7))
connections[-8] = -7
# run GGA Band structure
spec = {'task_type': '{} band structure'.format(type_name), '_queueadapter': QA_VASP,
'_dupefinder': DupeFinderVasp().to_dict(), '_priority': priority}
fws.append(FireWork([VaspCopyTask({'use_CONTCAR': False}), SetupNonSCFTask({'mode': 'line'}),
get_custodian_task(spec)], spec, name=get_slug(f+'--'+spec['task_type']),
fw_id=-6))
connections[-7] = -6
# insert into DB - GGA Band structure
spec = {'task_type': 'VASP db insertion', '_queueadapter': QA_DB,
'_allow_fizzled_parents': True, '_priority': priority}
fws.append(FireWork([VaspToDBTask({})], spec, name=get_slug(f+'--'+spec['task_type']), fw_id=-5))
connections[-6] = -5
wf = Workflow(fws, connections)
print 'Done adding more runs...'
return FWAction(additions=wf)
return FWAction()
vector = zeros((MAX_Z))
for element in composition:
fraction = composition.get_atomic_fraction(element)
vector[element.Z - 1] = fraction
return(vector)
# Extract materials and band gaps into lists, and construct naive feature set
materials = []
bandgaps = []
naiveFeatures = []
MAX_Z = 100 # maximum length of vector to hold naive feature set
for line in trainFile:
split = str.split(line, ',')
material = Composition(split[0])
materials.append(material) #store chemical formulas
naiveFeatures.append(naiveVectorize(material)) #create features from chemical formula
bandgaps.append(float(split[1])) #store numerical values of band gaps
##############################################################################################################
# Establish baseline accuracy by "guessing the average" of the band gap set
# A good model should never do worse.
baselineError = mean(abs(mean(bandgaps) - bandgaps))
print("The MAE of always guessing the average band gap is: " + str(round(baselineError, 3)) + " eV")
##############################################################################################################
#alpha is a tuning parameter affecting how regression deals with collinear inputs
linear = linear_model.Ridge(alpha = 0.5)
def run_task(self, fw_spec):
if '_fizzled_parents' in fw_spec and not 'prev_vasp_dir' in fw_spec:
prev_dir = get_loc(fw_spec['_fizzled_parents'][0]['launches'][0]['launch_dir'])
update_spec = {} # add this later when creating new FW
fizzled_parent = True
parse_dos = False
else:
prev_dir = get_loc(fw_spec['prev_vasp_dir'])
update_spec = {'prev_vasp_dir': prev_dir,
'prev_task_type': fw_spec['prev_task_type'],
'run_tags': fw_spec['run_tags'], 'parameters': fw_spec.get('parameters')}
fizzled_parent = False
parse_dos = 'Uniform' in fw_spec['prev_task_type']
if 'run_tags' in fw_spec:
self.additional_fields['run_tags'] = fw_spec['run_tags']
else:
self.additional_fields['run_tags'] = fw_spec['_fizzled_parents'][0]['spec']['run_tags']
if MOVE_TO_GARDEN_DEV:
prev_dir = move_to_garden(prev_dir, prod=False)
elif MOVE_TO_GARDEN_PROD:
prev_dir = move_to_garden(prev_dir, prod=True)
# get the directory containing the db file
db_dir = os.environ['DB_LOC']
db_path = os.path.join(db_dir, 'tasks_db.json')
logging.basicConfig(level=logging.INFO)
logger = logging.getLogger('MPVaspDrone')
logger.setLevel(logging.INFO)
sh = logging.StreamHandler(stream=sys.stdout)
sh.setLevel(getattr(logging, 'INFO'))
logger.addHandler(sh)
with open(db_path) as f:
db_creds = json.load(f)
drone = MPVaspDrone(
host=db_creds['host'], port=db_creds['port'],
database=db_creds['database'], user=db_creds['admin_user'],
password=db_creds['admin_password'],
collection=db_creds['collection'], parse_dos=parse_dos,
additional_fields=self.additional_fields,
update_duplicates=self.update_duplicates)
t_id, d = drone.assimilate(prev_dir, launches_coll=LaunchPad.auto_load().launches)
mpsnl = d['snl_final'] if 'snl_final' in d else d['snl']
snlgroup_id = d['snlgroup_id_final'] if 'snlgroup_id_final' in d else d['snlgroup_id']
update_spec.update({'mpsnl': mpsnl, 'snlgroup_id': snlgroup_id})
print 'ENTERED task id:', t_id
stored_data = {'task_id': t_id}
if d['state'] == 'successful':
update_spec['analysis'] = d['analysis']
update_spec['output'] = d['output']
return FWAction(stored_data=stored_data, update_spec=update_spec)
# not successful - first test to see if UnconvergedHandler is needed
if not fizzled_parent:
unconverged_tag = 'unconverged_handler--{}'.format(fw_spec['prev_task_type'])
output_dir = last_relax(os.path.join(prev_dir, 'vasprun.xml'))
ueh = UnconvergedErrorHandler(output_filename=output_dir)
if ueh.check() and unconverged_tag not in fw_spec['run_tags']:
print 'Unconverged run! Creating dynamic FW...'
spec = {'prev_vasp_dir': prev_dir,
'prev_task_type': fw_spec['task_type'],
'mpsnl': mpsnl, 'snlgroup_id': snlgroup_id,
'task_type': fw_spec['prev_task_type'],
'run_tags': list(fw_spec['run_tags']),
'parameters': fw_spec.get('parameters'),
'_dupefinder': DupeFinderVasp().to_dict(),
'_priority': fw_spec['_priority']}
snl = StructureNL.from_dict(spec['mpsnl'])
spec['run_tags'].append(unconverged_tag)
spec['_queueadapter'] = QA_VASP
fws = []
connections = {}
f = Composition.from_formula(
snl.structure.composition.reduced_formula).alphabetical_formula
fws.append(FireWork(
[VaspCopyTask({'files': ['INCAR', 'KPOINTS', 'POSCAR', 'POTCAR', 'CONTCAR'],
'use_CONTCAR': False}), SetupUnconvergedHandlerTask(),
get_custodian_task(spec)], spec, name=get_slug(f + '--' + spec['task_type']),
fw_id=-2))
spec = {'task_type': 'VASP db insertion', '_allow_fizzled_parents': True,
'_priority': fw_spec['_priority'], '_queueadapter': QA_DB,
'run_tags': list(fw_spec['run_tags'])}
spec['run_tags'].append(unconverged_tag)
fws.append(
FireWork([VaspToDBTask()], spec, name=get_slug(f + '--' + spec['task_type']),
fw_id=-1))
connections[-2] = -1
wf = Workflow(fws, connections)
return FWAction(detours=wf)
# not successful and not due to convergence problem - FIZZLE
raise ValueError("DB insertion successful, but don't know how to fix this FireWork! Can't continue with workflow...")
