def test_no_scaling(self):
sm = StructureMatcher(ltol=0.1, stol=0.1, angle_tol=2,
scale=False, comparator=ElementComparator())
self.assertTrue(sm.fit(self.struct_list[0], self.struct_list[1]))
self.assertTrue(sm.get_rms_dist(self.struct_list[0],
self.struct_list[1])[0] < 0.0008)
def get_framework_rms_plot(self,
plt=None,
granularity=200,
matching_s=None):
"""
Get the plot of rms framework displacement vs time. Useful for checking
for melting, especially if framework atoms can move via paddle-wheel
or similar mechanism (which would show up in max framework displacement
but doesn't constitute melting).
Args:
plt (matplotlib.pyplot): If plt is supplied, changes will be made
to an existing plot. Otherwise, a new plot will be created.
granularity (int): Number of structures to match
matching_s (Structure): Optionally match to a disordered structure
instead of the first structure in the analyzer. Required when
a secondary mobile ion is present.
Notes:
The method doesn't apply to NPT-AIMD simulation analysis.
"""
from pymatgen.util.plotting import pretty_plot
if self.lattices is not None and len(self.lattices) > 1:
warnings.warn(
"Note the method doesn't apply to NPT-AIMD simulation analysis!"
)
plt = pretty_plot(12, 8, plt=plt)
step = (self.corrected_displacements.shape[1] - 1) // (granularity - 1)
f = (matching_s or self.structure).copy()
f.remove_species([self.specie])
sm = StructureMatcher(
primitive_cell=False,
stol=0.6,
comparator=OrderDisorderElementComparator(),
allow_subset=True,
)
rms = []
for s in self.get_drift_corrected_structures(step=step):
s.remove_species([self.specie])
d = sm.get_rms_dist(f, s)
if d:
rms.append(d)
else:
rms.append((1, 1))
max_dt = (len(rms) - 1) * step * self.step_skip * self.time_step
if max_dt > 100000:
plot_dt = np.linspace(0, max_dt / 1000, len(rms))
unit = "ps"
else:
plot_dt = np.linspace(0, max_dt, len(rms))
unit = "fs"
rms = np.array(rms)
plt.plot(plot_dt, rms[:, 0], label="RMS")
plt.plot(plot_dt, rms[:, 1], label="max")
plt.legend(loc="best")
plt.xlabel(f"Timestep ({unit})")
plt.ylabel("normalized distance")
plt.tight_layout()
return plt
def test_no_scaling(self):
sm = StructureMatcher(ltol=0.1, stol=0.1, angle_tol=2,
scale=False, comparator=ElementComparator())
self.assertTrue(sm.fit(self.struct_list[0], self.struct_list[1]))
self.assertTrue(sm.get_rms_dist(self.struct_list[0],
self.struct_list[1])[0] < 0.0008)
def get_framework_rms_plot(self, plt=None, granularity=200,
matching_s=None):
"""
Get the plot of rms framework displacement vs time. Useful for checking
for melting, especially if framework atoms can move via paddle-wheel
or similar mechanism (which would show up in max framework displacement
but doesn't constitute melting).
Args:
plt (matplotlib.pyplot): If plt is supplied, changes will be made
to an existing plot. Otherwise, a new plot will be created.
granularity (int): Number of structures to match
matching_s (Structure): Optionally match to a disordered structure
instead of the first structure in the analyzer. Required when
a secondary mobile ion is present.
Notes:
The method doesn't apply to NPT-AIMD simulation analysis.
"""
from pymatgen.util.plotting import pretty_plot
if self.lattices is not None and len(self.lattices) > 1:
warnings.warn("Note the method doesn't apply to NPT-AIMD "
"simulation analysis!")
plt = pretty_plot(12, 8, plt=plt)
step = (self.corrected_displacements.shape[1] - 1) // (granularity - 1)
f = (matching_s or self.structure).copy()
f.remove_species([self.specie])
sm = StructureMatcher(primitive_cell=False, stol=0.6,
comparator=OrderDisorderElementComparator(),
allow_subset=True)
rms = []
for s in self.get_drift_corrected_structures(step=step):
s.remove_species([self.specie])
d = sm.get_rms_dist(f, s)
if d:
rms.append(d)
else:
rms.append((1, 1))
max_dt = (len(rms) - 1) * step * self.step_skip * self.time_step
if max_dt > 100000:
plot_dt = np.linspace(0, max_dt / 1000, len(rms))
unit = 'ps'
else:
plot_dt = np.linspace(0, max_dt, len(rms))
unit = 'fs'
rms = np.array(rms)
plt.plot(plot_dt, rms[:, 0], label='RMS')
plt.plot(plot_dt, rms[:, 1], label='max')
plt.legend(loc='best')
plt.xlabel("Timestep ({})".format(unit))
plt.ylabel("normalized distance")
plt.tight_layout()
return plt
def test_rms_vs_minimax(self):
# This tests that structures with adjusted RMS less than stol, but minimax
# greater than stol are treated properly
sm = StructureMatcher(ltol=0.2, stol=0.3, angle_tol=5, primitive_cell=False)
l = Lattice.orthorhombic(1, 2, 12)
sp = ["Si", "Si", "Al"]
s1 = Structure(l, sp, [[0.5, 0, 0], [0, 0, 0], [0, 0, 0.5]])
s2 = Structure(l, sp, [[0.5, 0, 0], [0, 0, 0], [0, 0, 0.6]])
self.assertArrayAlmostEqual(sm.get_rms_dist(s1, s2), (0.32 ** 0.5 / 2, 0.4))
self.assertEqual(sm.fit(s1, s2), False)
self.assertEqual(sm.fit_anonymous(s1, s2), False)
self.assertEqual(sm.get_mapping(s1, s2), None)
def get_framework_rms_plot(self,
plt=None,
granularity=200,
matching_s=None):
"""
Get the plot of rms framework displacement vs time. Useful for checking
for melting, especially if framework atoms can move via paddle-wheel
or similar mechanism (which would show up in max framework displacement
but doesn't constitute melting).
Args:
granularity (int): Number of structures to match
matching_s (Structure): Optionally match to a disordered structure
instead of the first structure in the analyzer. Required when
a secondary mobile ion is present.
"""
from pymatgen.util.plotting_utils import get_publication_quality_plot
plt = get_publication_quality_plot(12, 8, plt=plt)
step = (self.corrected_displacements.shape[1] - 1) // (granularity - 1)
f = (matching_s or self.structure).copy()
f.remove_species([self.specie])
sm = StructureMatcher(primitive_cell=False,
stol=0.6,
comparator=OrderDisorderElementComparator(),
allow_subset=True)
rms = []
for s in self.get_drift_corrected_structures(step=step):
s.remove_species([self.specie])
d = sm.get_rms_dist(f, s)
if d:
rms.append(d)
else:
rms.append((1, 1))
max_dt = (len(rms) - 1) * step * self.step_skip * self.time_step
if max_dt > 100000:
plot_dt = np.linspace(0, max_dt / 1000, len(rms))
unit = 'ps'
else:
plot_dt = np.linspace(0, max_dt, len(rms))
unit = 'fs'
rms = np.array(rms)
plt.plot(plot_dt, rms[:, 0], label='RMS')
plt.plot(plot_dt, rms[:, 1], label='max')
plt.legend(loc='best')
plt.xlabel("Timestep ({})".format(unit))
plt.ylabel("normalized distance")
plt.tight_layout()
return plt
def test_rms_vs_minimax(self):
# This tests that structures with adjusted RMS less than stol, but minimax
# greater than stol are treated properly
# stol=0.3 gives exactly an ftol of 0.1 on the c axis
sm = StructureMatcher(ltol=0.2, stol=0.301, angle_tol=1, primitive_cell=False)
l = Lattice.orthorhombic(1, 2, 12)
sp = ["Si", "Si", "Al"]
s1 = Structure(l, sp, [[0.5, 0, 0], [0, 0, 0], [0, 0, 0.5]])
s2 = Structure(l, sp, [[0.5, 0, 0], [0, 0, 0], [0, 0, 0.6]])
self.assertArrayAlmostEqual(sm.get_rms_dist(s1, s2),
(0.32 ** 0.5 / 2, 0.4))
self.assertEqual(sm.fit(s1, s2), False)
self.assertEqual(sm.fit_anonymous(s1, s2), False)
self.assertEqual(sm.get_mapping(s1, s2), None)
def post_process(self, docs):
s1 = Structure.from_dict(self.structure)
m = StructureMatcher(
ltol=self.ltol,
stol=self.stol,
angle_tol=self.angle_tol,
primitive_cell=True,
scale=True,
attempt_supercell=False,
comparator=ElementComparator(),
)
matches = []
for doc in docs:
s2 = Structure.from_dict(doc["structure"])
matched = m.fit(s1, s2)
if matched:
rms = m.get_rms_dist(s1, s2)
matches.append({
"material_id": doc["material_id"],
"normalized_rms_displacement": rms[0],
"max_distance_paired_sites": rms[1],
})
response = sorted(
matches[:self.limit],
key=lambda x: (
x["normalized_rms_displacement"],
x["max_distance_paired_sites"],
),
)
return response
def test_supercell_subsets(self):
sm = StructureMatcher(ltol=0.2, stol=0.3, angle_tol=5,
primitive_cell=False, scale=True,
attempt_supercell=True, allow_subset=True,
supercell_size='volume')
sm_no_s = StructureMatcher(ltol=0.2, stol=0.3, angle_tol=5,
primitive_cell=False, scale=True,
attempt_supercell=True, allow_subset=False,
supercell_size='volume')
l = Lattice.orthorhombic(1, 2, 3)
s1 = Structure(l, ['Ag', 'Si', 'Si'],
[[.7,.4,.5],[0,0,0.1],[0,0,0.2]])
s1.make_supercell([2,1,1])
s2 = Structure(l, ['Si', 'Si', 'Ag'],
[[0,0.1,-0.95],[0,0.1,0],[-.7,.5,.375]])
shuffle = [0,2,1,3,4,5]
s1 = Structure.from_sites([s1[i] for i in shuffle])
#test when s1 is exact supercell of s2
result = sm.get_s2_like_s1(s1, s2)
for a, b in zip(s1, result):
self.assertTrue(a.distance(b) < 0.08)
self.assertEqual(a.species_and_occu, b.species_and_occu)
self.assertTrue(sm.fit(s1, s2))
self.assertTrue(sm.fit(s2, s1))
self.assertTrue(sm_no_s.fit(s1, s2))
self.assertTrue(sm_no_s.fit(s2, s1))
rms = (0.048604032430991401, 0.059527539448807391)
self.assertTrue(np.allclose(sm.get_rms_dist(s1, s2), rms))
self.assertTrue(np.allclose(sm.get_rms_dist(s2, s1), rms))
#test when the supercell is a subset of s2
subset_supercell = s1.copy()
del subset_supercell[0]
result = sm.get_s2_like_s1(subset_supercell, s2)
self.assertEqual(len(result), 6)
for a, b in zip(subset_supercell, result):
self.assertTrue(a.distance(b) < 0.08)
self.assertEqual(a.species_and_occu, b.species_and_occu)
self.assertTrue(sm.fit(subset_supercell, s2))
self.assertTrue(sm.fit(s2, subset_supercell))
self.assertFalse(sm_no_s.fit(subset_supercell, s2))
self.assertFalse(sm_no_s.fit(s2, subset_supercell))
rms = (0.053243049896333279, 0.059527539448807336)
self.assertTrue(np.allclose(sm.get_rms_dist(subset_supercell, s2), rms))
self.assertTrue(np.allclose(sm.get_rms_dist(s2, subset_supercell), rms))
#test when s2 (once made a supercell) is a subset of s1
s2_missing_site = s2.copy()
del s2_missing_site[1]
result = sm.get_s2_like_s1(s1, s2_missing_site)
for a, b in zip((s1[i] for i in (0, 2, 4, 5)), result):
self.assertTrue(a.distance(b) < 0.08)
self.assertEqual(a.species_and_occu, b.species_and_occu)
self.assertTrue(sm.fit(s1, s2_missing_site))
self.assertTrue(sm.fit(s2_missing_site, s1))
self.assertFalse(sm_no_s.fit(s1, s2_missing_site))
self.assertFalse(sm_no_s.fit(s2_missing_site, s1))
rms = (0.029763769724403633, 0.029763769724403987)
self.assertTrue(np.allclose(sm.get_rms_dist(s1, s2_missing_site), rms))
self.assertTrue(np.allclose(sm.get_rms_dist(s2_missing_site, s1), rms))
def test_supercell_subsets(self):
sm = StructureMatcher(ltol=0.2,
stol=0.3,
angle_tol=5,
primitive_cell=False,
scale=True,
attempt_supercell=True,
allow_subset=True,
supercell_size='volume')
sm_no_s = StructureMatcher(ltol=0.2,
stol=0.3,
angle_tol=5,
primitive_cell=False,
scale=True,
attempt_supercell=True,
allow_subset=False,
supercell_size='volume')
l = Lattice.orthorhombic(1, 2, 3)
s1 = Structure(l, ['Ag', 'Si', 'Si'],
[[.7, .4, .5], [0, 0, 0.1], [0, 0, 0.2]])
s1.make_supercell([2, 1, 1])
s2 = Structure(l, ['Si', 'Si', 'Ag'],
[[0, 0.1, -0.95], [0, 0.1, 0], [-.7, .5, .375]])
shuffle = [0, 2, 1, 3, 4, 5]
s1 = Structure.from_sites([s1[i] for i in shuffle])
#test when s1 is exact supercell of s2
result = sm.get_s2_like_s1(s1, s2)
for a, b in zip(s1, result):
self.assertTrue(a.distance(b) < 0.08)
self.assertEqual(a.species_and_occu, b.species_and_occu)
self.assertTrue(sm.fit(s1, s2))
self.assertTrue(sm.fit(s2, s1))
self.assertTrue(sm_no_s.fit(s1, s2))
self.assertTrue(sm_no_s.fit(s2, s1))
rms = (0.048604032430991401, 0.059527539448807391)
self.assertTrue(np.allclose(sm.get_rms_dist(s1, s2), rms))
self.assertTrue(np.allclose(sm.get_rms_dist(s2, s1), rms))
#test when the supercell is a subset of s2
subset_supercell = s1.copy()
del subset_supercell[0]
result = sm.get_s2_like_s1(subset_supercell, s2)
self.assertEqual(len(result), 6)
for a, b in zip(subset_supercell, result):
self.assertTrue(a.distance(b) < 0.08)
self.assertEqual(a.species_and_occu, b.species_and_occu)
self.assertTrue(sm.fit(subset_supercell, s2))
self.assertTrue(sm.fit(s2, subset_supercell))
self.assertFalse(sm_no_s.fit(subset_supercell, s2))
self.assertFalse(sm_no_s.fit(s2, subset_supercell))
rms = (0.053243049896333279, 0.059527539448807336)
self.assertTrue(np.allclose(sm.get_rms_dist(subset_supercell, s2),
rms))
self.assertTrue(np.allclose(sm.get_rms_dist(s2, subset_supercell),
rms))
#test when s2 (once made a supercell) is a subset of s1
s2_missing_site = s2.copy()
del s2_missing_site[1]
result = sm.get_s2_like_s1(s1, s2_missing_site)
for a, b in zip((s1[i] for i in (0, 2, 4, 5)), result):
self.assertTrue(a.distance(b) < 0.08)
self.assertEqual(a.species_and_occu, b.species_and_occu)
self.assertTrue(sm.fit(s1, s2_missing_site))
self.assertTrue(sm.fit(s2_missing_site, s1))
self.assertFalse(sm_no_s.fit(s1, s2_missing_site))
self.assertFalse(sm_no_s.fit(s2_missing_site, s1))
rms = (0.029763769724403633, 0.029763769724403987)
self.assertTrue(np.allclose(sm.get_rms_dist(s1, s2_missing_site), rms))
self.assertTrue(np.allclose(sm.get_rms_dist(s2_missing_site, s1), rms))
from pymatgen import Structure
from pymatgen.analysis.structure_matcher import StructureMatcher
client = pymongo.MongoClient()
db = client.springer
coll = db['pauling_file_unique_Parse']
if __name__ == '__main__':
for doc in coll.find({'key': 'sd_1223808'}):
struc1 = Structure.from_dict(doc['structure'])
for doc in coll.find({'key': 'sd_0458111'}):
struc2 = Structure.from_dict(doc['structure'])
for doc in coll.find({'key': 'sd_1933177'}):
struc3 = Structure.from_dict(doc['structure'])
for doc in coll.find({'key': 'sd_1010018'}):
struc4 = Structure.from_dict(doc['structure'])
print struc4
print doc['metadata']['_Springer']['geninfo']['Phase Label(s)']
# print Structure.from_dict(doc['structure'])
# print doc['structure']
for doc in coll.find({'key': 'sd_0529813'}):
struc5 = Structure.from_dict(doc['structure'])
matcher = StructureMatcher()
print matcher.fit(struc1, struc2), '8.18, 8.26', matcher.get_rms_dist(struc1, struc2)
print matcher.fit(struc2, struc3), '8.26 8.25', matcher.get_rms_dist(struc2, struc3)
print matcher.fit(struc3, struc1), '8.25 8.18', matcher.get_rms_dist(struc3, struc1)
print matcher.fit(struc1, struc4), matcher.get_rms_dist(struc1, struc4)
print matcher.fit(struc1, struc5), matcher.get_rms_dist(struc1, struc5)
print matcher.fit(struc2, struc5), matcher.get_rms_dist(struc2, struc5)
async def find_structure(
structure: Structure = Body(
...,
title="Pymatgen structure object to query with",
),
ltol: float = Query(
0.2,
title="Fractional length tolerance. Default is 0.2.",
),
stol: float = Query(
0.3,
title=
"Site tolerance. Defined as the fraction of the average free \
length per atom := ( V / Nsites ) ** (1/3). Default is 0.3.",
),
angle_tol: float = Query(
5,
title="Angle tolerance in degrees. Default is 5 degrees.",
),
limit: int = Query(
1,
title=
"Maximum number of matches to show. Defaults to 1, only showing the best match.",
),
):
"""
Obtains material structures that match a given input structure within some tolerance.
Returns:
A list of Material IDs for materials with matched structures alongside the associated RMS values
"""
try:
s = PS.from_dict(structure.dict())
except Exception:
raise HTTPException(
status_code=404,
detail=
"Body cannot be converted to a pymatgen structure object.",
)
m = StructureMatcher(
ltol=ltol,
stol=stol,
angle_tol=angle_tol,
primitive_cell=True,
scale=True,
attempt_supercell=False,
comparator=ElementComparator(),
)
crit = {"composition_reduced": dict(s.composition.to_reduced_dict)}
self.store.connect()
matches = []
for r in self.store.query(criteria=crit,
properties=["structure", "task_id"]):
s2 = PS.from_dict(r["structure"])
matched = m.fit(s, s2)
if matched:
rms = m.get_rms_dist(s, s2)
matches.append({
"task_id": r["task_id"],
"normalized_rms_displacement": rms[0],
"max_distance_paired_sites": rms[1],
})
response = {
"data":
sorted(
matches[:limit],
key=lambda x: (
x["normalized_rms_displacement"],
x["max_distance_paired_sites"],
),
)
}
return response
import pymatgen as mg
from pymatgen.analysis.structure_matcher import StructureMatcher
struct1 = mg.Structure.from_file("1.cif")
struct2 = mg.Structure.from_file("2.cif")
matcher = StructureMatcher(attempt_supercell=True, primitive_cell=False)
rms = matcher.get_rms_dist(struct1, struct2)
print rms
