def sa(al_comp, kwargs, eci, db_name, size, lattice):
comp = {"Al": al_comp, "Zn": 1.0 - al_comp}
bc = BulkCrystal(**kwargs)
calc = get_ce_calc(bc, kwargs, eci, size)
bc = calc.BC
bc.atoms.set_calculator(calc)
temperatures = [800, 700, 600, 500, 400, 300, 200, 100]
N = len(bc.atoms)
if rank == 0:
print("Supercell has {} atoms".format(N))
# Define parameters for equillibration
equil_params = {"maxiter": 10 * N, "mode": "fixed"}
nsteps = 100 * N
calc.set_composition(comp)
for T in temperatures:
mc = Montecarlo(bc.atoms, T, mpicomm=comm)
mc.runMC(mode="fixed", steps=nsteps, equil_params=equil_params)
thermo = mc.get_thermodynamic()
thermo["converged"] = True
thermo["al_conc"] = al_comp
thermo["temperature"] = T
if rank == 0:
db = dataset.connect("sqlite:///{}".format(db_name))
tbl = db["results"]
tbl.insert(thermo)
if rank == 0:
fname = "data/atoms_{}_{}.xyz".format(lattice,
bc.atoms.get_chemical_formula())
write(fname, bc.atoms)
def test_no_throw(self):
if not available:
self.skipTest(skipMsg)
no_throw = True
msg = ""
try:
bc = get_ternary_BC()
eci = get_example_ecis(bc)
atoms = bc.atoms.copy()
CE(atoms, bc, eci=eci)
mc = Montecarlo(atoms, 1000)
mc.insert_symbol_random_places("Mg", num=10, swap_symbs=["Al"])
# Note: convergence_factor should never be
# set to a negative value! Here it is done
# to make the test converged after one step
bias = AdaptiveBiasReactionPathSampler(
mc_obj=mc,
react_crd_init=DummyReacCrdInit(),
react_crd=[-1.0, 1.0],
convergence_factor=-1.0)
bias.run()
except Exception as exc:
msg = str(exc)
no_throw = False
self.assertTrue(no_throw, msg=msg)
def run(T, mg_conc, si_conc, precs):
conc_args = {
"conc_ratio_min_1": [[64, 0, 0]],
"conc_ratio_max_1": [[24, 40, 0]],
"conc_ratio_min_2": [[64, 0, 0]],
"conc_ratio_max_2": [[22, 21, 21]]
}
orig_spin_dict = {
"Mg": 1.0,
"Si": -1.0,
"Al": 0.0
}
kwargs = {
"crystalstructure": "fcc",
"a": 4.05,
"size": [4, 4, 4],
"basis_elements": [["Mg", "Si", "Al"]],
"conc_args": conc_args,
"db_name": "data/almgsi.db",
"max_cluster_size": 4
}
ceBulk = BulkCrystal(**kwargs)
ceBulk.spin_dict = orig_spin_dict
ceBulk.basis_functions = ceBulk._get_basis_functions()
ceBulk._get_cluster_information()
eci_file = "data/almgsi_fcc_eci.json"
with open(eci_file, 'r') as infile:
ecis = json.load(infile)
print(ecis)
#calc = CE( ceBulk, ecis, size=(3,3,3) )
calc = get_ce_calc(ceBulk, kwargs, ecis, size=[10, 10, 10])
ceBulk = calc.BC
ceBulk.atoms.set_calculator(calc)
comp = {
"Mg": mg_conc,
"Si": si_conc,
"Al": 1.0 - mg_conc - si_conc
}
calc.set_composition(comp)
for temp, prec in zip(T, precs):
print("Current temperature {}K".format(temp))
mc_obj = Montecarlo(ceBulk.atoms, temp, mpicomm=comm)
mode = "prec"
mc_obj.runMC(mode=mode, prec=prec)
thermo = mc_obj.get_thermodynamic()
thermo["temperature"] = temp
thermo["prec"] = prec
thermo["internal_energy"] = thermo.pop("energy")
thermo["converged"] = True
thermo["prec"] = prec
if (rank == 0):
db = dataset.connect("sqlite:///{}".format(mc_db_name))
tbl = db["results"]
thermo["sysID"] = sysID
tbl.insert(thermo)
def main():
energies = []
sizes = list(range(1, 6))
atoms = get_atoms()
two_atoms(atoms)
exit()
insert_np(6, atoms)
mc = Montecarlo(atoms, 0.1)
camera = Snapshot(atoms=mc.atoms, trajfile="/work/sophus/nuc_cluster.traj")
db = dataset.connect("sqlite:////work/sophus/mgsi_nuc_barrier_kamijo.db")
tbl = db["systems"]
mc.attach(camera, interval=100 * len(atoms))
num_mg = sum(1 for atom in atoms if atom.symbol == "Mg")
while num_mg > 2:
print(atoms.get_chemical_formula())
mc.runMC(mode="fixed", equil=False, steps=100 * len(atoms))
thermo = mc.get_thermodynamic()
tbl.insert(thermo)
# Remove one Mg atom and one Si atom
symbols = [a.symbol for a in atoms]
for i in range(20):
i = symbols.index("Si")
symbols[i] = "Al"
i = symbols.index("Mg")
symbols[i] = "Al"
mc.set_symbols(symbols)
num_mg = sum(1 for atom in atoms if atom.symbol == "Mg")
def pure_phase_entropy(small_bc):
from cemc.mcmc import Montecarlo
from ase.clease.tools import wrap_and_sort_by_position
from ase.io import read
from cemc.mcmc import SiteOrderParameter, Snapshot
import dataset
T = [1, 5, 10, 20, 30, 40, 50, 75, 100, 150, 200, 210, 220, 230, 240, 250, 260, 270, 280, 290, 300, 310,
320, 330, 340, 350, 360, 370, 380, 390, 400, 420, 440, 460, 480, 500]
calc = get_ce_calc(small_bc, kwargs, ecis, size=[10, 10, 10], db_name="data/db10x10x10Al3Mg.db")
bc = calc.BC
bc.atoms.set_calculator(calc)
atoms = read("data/al3mg_template.xyz")
atoms = wrap_and_sort_by_position(atoms)
symbs = [atom.symbol for atom in atoms]
calc.set_symbols(symbs)
site_order = SiteOrderParameter(bc.atoms)
db = dataset.connect("sqlite:///data/pure_al3mg3.db")
syst = db["thermodynamic"]
camera = Snapshot(trajfile="data/pure_phase_entropy.traj", atoms=bc.atoms)
for temp in T:
print("Current temperature: {}K".format(temp))
site_order.reset()
mc = Montecarlo(bc.atoms, temp)
mc.attach(site_order)
mc.attach(camera, interval=50000)
equil_param = {"window_length": 100000, "mode": "fixed"}
mc.runMC(mode="fixed", steps=500000, equil=True, equil_params=equil_param)
mean, stddev = site_order.get_average()
thermo = mc.get_thermodynamic()
thermo["site_order"] = mean
thermo["site_order_std"] = stddev
syst.insert(thermo)
def _estimate_internal_energy(self, conc, sweeps=2):
"""
Estimates the internal energy of one structure
"""
try:
self.atoms._calc.set_composition(conc)
mc = Montecarlo(self.atoms, self.temperature)
mc.runMC(mode="fixed", steps=sweeps * len(self.atoms), equil=False)
energy = mc.get_thermodynamic()["energy"]
except TooFewElementsError as exc:
energy = 1.0
return energy
def test_no_throw(self):
if not available:
self.skipTest(import_msg)
ceBulk, atoms = self.init_bulk_crystal()
mc = Montecarlo(atoms, 100.0)
mc.insert_symbol_random_places("Mg", num=5, swap_symbs=["Al"])
mc.insert_symbol_random_places("Si", num=5, swap_symbs=["Al"])
par_temp = ParallelTempering(mc_obj=mc, Tmax=100.0, Tmin=0.001)
mc_args = {"steps": 100, "equil": False}
par_temp.run(mc_args=mc_args, num_exchange_cycles=3)
os.remove("temp_scheme.csv")
def test_run(self):
if not available:
self.skipTest(reason)
bc = get_ternary_BC()
eci = get_example_ecis(bc)
atoms = bc.atoms.copy()
calc = CE(atoms, bc, eci=eci)
mc = Montecarlo(atoms, 1000000)
mc.insert_symbol_random_places("Mg", swap_symbs=["Al"], num=3)
performance_monitor = MultithreadPerformance(4)
performance_monitor.run(mc, 100)
def main(size, T):
atoms = create_surface(size)
atoms = wrap_and_sort_by_position(atoms)
conc_args = {
"conc_ratio_min_1": [[64, 0, 0]],
"conc_ratio_max_1": [[24, 40, 0]],
"conc_ratio_min_2": [[64, 0, 0]],
"conc_ratio_max_2": [[22, 21, 21]]
}
kwargs = {
"crystalstructure": "fcc",
"a": 4.05,
"size": [4, 4, 4],
"basis_elements": [["Al", "Mg", "Si"]],
"conc_args": conc_args,
"db_name": "data/almgsi.db",
"max_cluster_size": 4
}
ceBulk = BulkCrystal(**kwargs)
eci_file = "data/almgsi_fcc_eci_newconfig.json"
with open(eci_file, 'r') as infile:
ecis = json.load(infile)
db_name = "large_cell_db{}x{}x{}.db".format(size[0], size[1], size[2])
calc = get_ce_calc(ceBulk, kwargs, ecis, size=size, db_name=db_name)
ceBulk = calc.BC
ceBulk.atoms.set_calculator(calc)
mc = Montecarlo(ceBulk.atoms, T)
symbs = [atom.symbol for atom in atoms]
mc.set_symbols(symbs)
# Write a copy of the current atoms object
from ase.io import write
shape_str = "-".join((str(item) for item in size))
uid_str = "{}K_{}".format(T, shape_str)
write(WORKDIR + "initial_config{}.xyz".format(uid_str), mc.atoms)
backup = MCBackup(mc,
backup_file=WORKDIR + "backup{}.xyz".format(uid_str),
db_name=WORKDIR + "mc_surface.db")
camera = Snapshot(atoms=mc.atoms,
trajfile=WORKDIR + "surface{}.traj".format(uid_str))
evol = EnergyEvolution(mc)
nsteps = int(10E6)
mc.attach(backup, interval=100000)
mc.attach(camera, interval=int(nsteps / 20))
mc.attach(evol, interval=10 * len(mc.atoms))
mc.runMC(mode="fixed", steps=nsteps, equil=False)
write(WORKDIR + "final_config{}.xyz".format(uid_str), mc.atoms)
def run(maxT, minT, n_temp, mg_conc):
T = np.linspace(minT, maxT, n_temp)[::-1]
conc_args = {
"conc_ratio_min_1": [[1, 0]],
"conc_ratio_max_1": [[0, 1]],
}
kwargs = {
"crystalstructure": "fcc",
"a": 4.05,
"size": [4, 4, 4],
"basis_elements": [["Al", "Mg"]],
"conc_args": conc_args,
"db_name": "data/temporary_bcdb.db",
"max_cluster_size": 4
}
ceBulk = BulkCrystal(**kwargs)
print(ceBulk.basis_functions)
eci_file = "data/ce_hydrostatic.json"
with open(eci_file, 'r') as infile:
ecis = json.load(infile)
print(ecis)
#calc = CE( ceBulk, ecis, size=(3,3,3) )
calc = get_ce_calc(ceBulk, kwargs, ecis, size=[10, 10, 10])
ceBulk = calc.BC
ceBulk.atoms.set_calculator(calc)
comp = {"Al": 1.0 - mg_conc, "Mg": mg_conc}
calc.set_composition(comp)
print("Number of atoms: {}".format(len(ceBulk.atoms)))
for temp in T:
print("Current temperature {}K".format(temp))
mc_obj = Montecarlo(ceBulk.atoms, temp, mpicomm=comm)
mode = "prec"
prec = 1E-5
mc_obj.runMC(mode=mode, prec=prec)
thermo = mc_obj.get_thermodynamic()
thermo["temperature"] = temp
thermo["prec"] = prec
thermo["internal_energy"] = thermo.pop("energy")
thermo["converged"] = True
if (rank == 0):
db = connect(mc_db_name)
db.write(ceBulk.atoms, key_value_pairs=thermo)
def test_no_throw(self):
if not available:
self.skipTest(import_msg)
msg = ""
no_throw = True
try:
ceBulk = self.init_bulk_crystal()
mc = Montecarlo(ceBulk.atoms, 100.0)
mc.insert_symbol_random_places("Mg", num=5, swap_symbs=["Al"])
mc.insert_symbol_random_places("Si", num=5, swap_symbs=["Al"])
par_temp = ParallelTempering(mc_obj=mc, Tmax=100.0, Tmin=0.001)
mc_args = {"steps": 100, "equil": False}
par_temp.run(mc_args=mc_args, num_exchange_cycles=3)
os.remove("temp_scheme.csv")
except Exception as exc:
msg = "{}: {}".format(type(exc).__name__, str(exc))
no_throw = False
self.assertTrue(no_throw, msg=msg)
def perform_mc(uid):
from ase.io import read
from cemc.mcmc import Montecarlo
db = dataset.connect(PHASE_DIAG_DB)
tbl = db["simulation_plan"]
row = tbl.find_one(id=uid)
print(row)
temperatures = list(np.arange(1, 600, 50))
ceBulk = get_ce_with_calc()
swap_symbs = row["swap_new"].split("-")
current = 0
if row["phase"] == "mgsi":
atoms = read("data/ground_stateMgSi.xyz")
symbols = [atom.symbol for atom in atoms]
else:
symbols = ["Al" for _ in range(len(ceBulk.atoms))]
num_inserted = np.zeros(len(swap_symbs))
for i in range(len(symbols)):
if symbols[i] == row["swap_old"]:
symbols[i] = swap_symbs[current]
num_inserted[current] += 1
if num_inserted[-1] >= row["num_insert"]:
break
elif num_inserted[current] >= row["num_insert"]:
current += 1
elif num_inserted[current] >= row["num_insert"]:
current += 1
ceBulk.atoms.get_calculator().set_symbols(symbols)
result_tab = db["simulations"]
for T in temperatures:
print("Current temperature: {}".format(T))
mc = Montecarlo(ceBulk.atoms, T)
equil_params = {"mode": "fixed", "maxiter": 10000}
mc.runMC(steps=1000 * len(symbols), equil_params=equil_params)
thermo = mc.get_thermodynamic()
thermo["runID"] = uid
result_tab.insert(thermo)
def get_thermodynamic(self):
"""
Override the thermodynamics function.
Normalization of the averagers:
The values in average track should represent the ratio
between partition functions when one have one extra of the
inserted atom and one less of the removed atom.
In the high temperature limit this is given by
Z_{n+1} = N!/((n+1)!(N-n-1)!)
and
Z_n = N!/(n!(N-n)!)
where N is the overall number of atoms of the two atoms being swapped.
Hence, the ratio becomes
Z_{n+1}/Z_n = (N-n)/(n+1),
the sums are normalized such that they results in this value
at high temperatures.
:return: dict with the thermodynamical properties
"""
self.collect_results()
res = {}
res = Montecarlo.get_thermodynamic(self)
at_count = self.count_atoms()
# Inser the insertion enegies
for move in self.insertion_moves:
key = self.get_key(move[0], move[1])
name = "insert_energy_{}".format(key)
N = self.averager_track[key].num_samples
inf_temp = float(at_count[move[0]]) / (at_count[move[1]] + 1)
E0 = self.averager_track[key].ref_value
res[name] = E0 - kB * self.T * \
np.log(self.averager_track[key].average * inf_temp)
name = "raw_insert_energy_{}".format(key)
res[name] = self.raw_insertion_energy[key] / N
name = "boltzmann_avg_insert_energy_{}".format(key)
res[name] = self.boltzmann_weight_ins_energy[key].average / \
self.averager_track[key].average
name = "boltzmann_avg_insert_energy_sq_{}".format(key)
res[name] = self.boltzmann_weight_ins_eng_sq[key].average / \
self.averager_track[key].average
return res
def _accept(self, system_changes):
"""Accept trial move.
:param list system_changes: Proposed changes
:return: True/False, if True the move is accepted
:rtype: bool
"""
move_accepted = Montecarlo._accept(self, system_changes)
if not self.move_ok():
return False
return move_accepted
def _get_trial_move(self):
"""Perform a trial move."""
if not self.nuc_sampler.is_in_window(self.network):
self.network(None)
msg = "System is outside the window before the trial move "
msg += "is performed!\n"
raise RuntimeError(msg)
#stat = self.network.get_statistics()
#num_clusters = stat["number_of_clusters"]
#max_size = stat["max_size"]
#lower,upper = self.nuc_sampler.get_window_boundaries(self.nuc_sampler.current_window)
#msg += "Num. clusters: {}. Max size: {}. Window limits: [{},{})".format(num_clusters,max_size,lower,upper)
return Montecarlo._get_trial_move(self)
def get_gs(self,
BC,
ecis=None,
composition=None,
temps=None,
n_steps_per_temp=1000,
atoms=None):
"""
Computes the ground states
:param BC: Instance of *CEBulk* or *CECrystal* from ASE
:param ecis: Dictionary with the Effecitve Cluster Interactions
:param composition: Dictionary with compositions (i.e. {"Mg":0.2,"Al":0.8})
:param temps: List of cooling temperatures
:param n_steps_per_temp: Number of MC steps per temperature
"""
if atoms is None:
atoms = BC.atoms.copy()
if atoms.get_calculator() is None:
if ecis is None:
raise ValueError("When a calculator is not attached "
"the ECIs has to be given!")
calc = CE(atoms, BC, ecis)
else:
calc = atoms.get_calculator()
#print (calc.get_cf())
if (temps is None):
temps = np.linspace(1, 1500, 30)[::-1]
if (composition is not None):
calc.set_composition(composition)
minimum_energy = LowestEnergyStructure(calc, None)
for T in temps:
print("Temperature {}".format(T))
mc_obj = Montecarlo(atoms, T)
mc_obj.constraints = self.constraints
minimum_energy.mc_obj = mc_obj
mc_obj.attach(minimum_energy)
mc_obj.runMC(steps=n_steps_per_temp, verbose=False, equil=False)
thermo = mc_obj.get_thermodynamic()
result = {
"atoms": minimum_energy.atoms,
"energy": minimum_energy.lowest_energy,
"cf": minimum_energy.lowest_energy_cf
}
return result
def sa(au_comp, kwargs, eci, db_name, size):
bc = CEBulk(**kwargs)
atoms = get_atoms_with_ce_calc(bc, kwargs, eci, size)
temperatures = [800, 700, 600, 500, 400, 300, 200, 100, 50, 25, 10]
N = len(bc.atoms)
# Define parameters for equillibration
equil_params = {
"maxiter": 10 * N,
"mode": "fixed"
}
comps = [0.05, 0.1, 0.15, 0.2, 0.25, 0.30, 0.35, 0.40, 0.45, 0.50, 0.55, 0.60, 0.65, 0.70, 0.75, 0.80, 0.85, 0.90, 0.95]
nsteps = 100 * N
calc = atoms.get_calculator()
for au_comp in comps:
comp = {
"Au": au_comp,
"Cu": 1.0-au_comp
}
calc.set_composition(comp)
for T in temperatures:
mc = Montecarlo(atoms, T)
mc.runMC(mode="fixed", steps=nsteps, equil_params=equil_params)
thermo = mc.get_thermodynamic()
thermo["converged"] = True
thermo["au_conc"] = au_comp
thermo["temperature"] = T
cf = calc.get_cf()
db = dataset.connect("sqlite:///{}".format(db_name))
tbl = db["results"]
uid = tbl.insert(thermo)
cf_tbl = db["corrfunc"]
cf["runID"] = uid
cf_tbl.insert(cf)
fname = "data/atoms_{}.xyz".format(atoms.get_chemical_formula())
write(fname, atoms)
def runMC(self, steps=100000, init_cluster=True, elements={}, equil=False):
"""
Run Monte Carlo for fixed nucleus size
:param int steps: Number of Monte Carlo steps
:param bool init_cluster: If True initialize a cluster, If False it is
assumed that a cluster of the correct size already exists in the
system
:param dict elements: Elements in the cluster
"""
if init_cluster:
self._check_nucleation_site_exists()
self.grow_cluster(elements)
self.network.collect_statistics = False
# Call one time
self.network([])
if self.network.num_root_nodes() > 1:
raise ValueError("Something went wrong during construction! "
"the system has more than one cluster!")
self.update_current_energy()
Montecarlo.runMC(self, steps=steps, equil=equil)
def run(N, T):
bc = init_bc(N)
mc = Montecarlo(bc.atoms, T)
nanop = get_nanoparticle()
symbs = insert_nano_particle(bc.atoms.copy(), nanop)
mc.set_symbols(symbs)
order_param = SiteOrderParameter(mc.atoms)
nsteps = int(1E6)
equil_params = {"mode": "fixed", "window_length": int(1E5)}
mc.attach(order_param)
mc.runMC(steps=nsteps, equil=True, equil_params=equil_params)
thermo = mc.get_thermodynamic()
mean, std = order_param.get_average()
thermo["order_param_mean"] = mean
thermo["order_param_std"] = std
thermo.update(equil_params)
db = dataset.connect("sqlite:///{}".format(mc_db_name))
tbl = db["cluster_stability"]
tbl.insert(thermo)
fname = workdir + "/final_structure{}K.xyz".format(T)
write(fname, mc.atoms)
def test_ignore_atoms(self):
if not has_ase_with_ce:
self.skipTest("ASE does not have CE")
from cemc.mcmc import FixedElement
no_trow = True
msg = ""
try:
from copy import deepcopy
conc = Concentration(basis_elements=[['V', 'Li'], ['O']])
kwargs = {
"crystalstructure": "rocksalt",
"concentration": conc,
"a": 4.12,
'size': [2, 2, 2],
'cubic': True,
"max_cluster_size": 4,
"max_cluster_dia": 4.12,
"db_name": 'database.db',
'basis_function': 'sluiter',
'ignore_background_atoms': True
}
fix_elem = FixedElement(element="O")
kw_args_cpy = deepcopy(kwargs)
ceBulk = CEBulk(**kw_args_cpy)
ecis = get_example_ecis(ceBulk)
atoms = get_atoms_with_ce_calc(ceBulk,
kwargs,
eci=ecis,
size=[3, 3, 3],
db_name="ignore_test_large.db")
calc = atoms.get_calculator()
# Insert some Li atoms
num_li = 5
symbols = [atom.symbol for atom in atoms]
num_inserted = 0
for i in range(0, len(symbols)):
if symbols[i] == "V":
symbols[i] = "Li"
num_inserted += 1
if num_inserted >= num_li:
break
calc.set_symbols(symbols)
mc = Montecarlo(atoms, 800)
mc.add_constraint(fix_elem)
mc.runMC(steps=100, equil=False, mode="fixed")
# Clean up files
os.remove("ignore_test_large.db")
os.remove("database.db")
except Exception as exc:
no_trow = False
msg = str(exc)
self.assertTrue(no_trow, msg=msg)
def plot_evolution(fname):
mc = Montecarlo.load(fname)
atoms_fname = fname.rpartition(".")[0]+"_final_atoms.xyz"
write(atoms_fname, mc.atoms)
energy_obs = None
for obs in mc.observers:
if obs[1].name == "EnergyEvolution":
energy_obs = obs[1]
if energy_obs is None:
raise RuntimeError("Did not find an energy evolution observer!")
energy = energy_obs.energies
fig = plt.figure()
ax = fig.add_subplot(1, 1, 1)
ax.plot(energy)
plt.show()
def _accept(self, system_changes):
"""Accept trial move.
:param list system_changes: Proposed changes
:return: True/False, if True the move is accepted
:rtype: bool
"""
# Note that we have to call the parent's accept first,
# as the _mc_step function assumes that an energy
# evaluation have been performed, prior to accepting
# or rejecting the move
move_accepted = Montecarlo._accept(self, system_changes)
if self.network.move_creates_new_cluster(system_changes):
return False
# if not self.move_ok():
# return False
return move_accepted
def sa(au_comp, kwargs, eci, db_name, size):
bc = CEBulk(**kwargs)
atoms = get_atoms_with_ce_calc(bc, kwargs, eci, size, db_name="cu-au_quad.db")
temperatures = [1500, 1400, 1300, 1200, 1100, 1000, 900, 800, 700, 600,
500, 400, 300, 200, 100, 50, 25, 10]
N = len(atoms)
gs = wrap_and_sort_by_position(read("data/atoms_Au250Cu750.xyz"))
symbs = [atom.symbol for atom in gs]
atoms.get_calculator().set_symbols(symbs)
print(atoms.get_calculator().get_energy())
exit()
# Define parameters for equillibration
equil_params = {
"maxiter": 10 * N,
"mode": "fixed"
}
nsteps = 200 * N
calc = atoms.get_calculator()
comp = {"Au": au_comp, "Cu": 1.0-au_comp}
calc.set_composition(comp)
energies = []
for T in temperatures:
mc = Montecarlo(atoms, T, accept_first_trial_move_after_reset=True)
energy_obs = EnergyEvolution(mc)
energies.append(energy_obs.energies)
mc.attach(energy_obs, interval=100)
mc.runMC(mode="fixed", steps=nsteps, equil_params=equil_params)
thermo = mc.get_thermodynamic()
thermo["converged"] = True
thermo["temperature"] = T
cf = calc.get_cf()
db = dataset.connect("sqlite:///{}".format(db_name))
tbl = db["results"]
uid = tbl.insert(thermo)
cf_tbl = db["corrfunc"]
cf["runID"] = uid
cf_tbl.insert(cf)
fname = "data/atoms_{}_{}.xyz".format(atoms.get_chemical_formula(), thermo['energy'])
write(fname, atoms)
np.savetxt("data/energy_evolution_{}.csv".format(atoms.get_chemical_formula()),
np.array(energies).T, delimiter=",")
def gs_mgsi():
atoms = get_atoms(cubic=True)
symbs = ["Mg" for _ in range(len(atoms))]
for i in range(int(len(symbs) / 2)):
symbs[i] = "Si"
atoms.get_calculator().set_symbols(symbs)
T = [
1500, 1400, 1300, 1200, 1100, 1000, 900, 800, 600, 500, 400, 300, 200,
100, 50
]
snap = Snapshot(trajfile="mgsi_gs_search.traj", atoms=atoms)
for temp in T:
print("Temperature {}K".format(temp))
mc = Montecarlo(atoms, temp)
mc.attach(snap, interval=10 * len(atoms))
mc.runMC(mode="fixed", steps=100 * len(atoms), equil=False)
def run_huge():
conc_args = {
"conc_ratio_min_1":[[64,0,0]],
"conc_ratio_max_1":[[24,40,0]],
"conc_ratio_min_2":[[64,0,0]],
"conc_ratio_max_2":[[22,21,21]]
}
kwargs = {
"crystalstructure":"fcc",
"size":[4,4,4],
"basis_elements":[["Al","Mg","Si"]],
"conc_args":conc_args,
"db_name":"some_db.db",
"max_cluster_size":4,
"a":4.05,
"ce_init_alg":"fast"
}
ceBulk = BulkCrystal( **kwargs )
with open(eci_file, 'r') as infile:
eci = json.load(infile)
calc = get_ce_calc(ceBulk, kwargs, eci=eci, size=[50,50,50])
calc.BC.atoms.set_calculator(calc)
print(calc.BC.basis_functions)
comp = {
"Al":0.9,
"Mg":0.05,
"Si":0.05
}
calc.set_composition(comp)
T = 150
T = [2000, 1500, 1200, 1000, 800, 600, 400, 293]
camera = Snapshot(trajfile=traj_file, atoms=calc.BC.atoms)
for temp in T:
print("Current temperature {}K".format(temp))
mc = Montecarlo(calc.BC.atoms, temp)
mc.attach(camera, interval=125000*20)
mc.runMC(mode="fixed", steps=125000*100, equil=False)
def run(T,mg_conc):
conc_args = {
"conc_ratio_min_1":[[1,0]],
"conc_ratio_max_1":[[0,1]],
}
kwargs = {
"crystalstructure":"fcc", "a":4.05, "size":[4,4,4], "basis_elements":[["Al","Mg"]],
"conc_args":conc_args, "db_name":"data/temporary_bcdb.db",
"max_cluster_size":4
}
ceBulk = BulkCrystal( **kwargs )
print (ceBulk.basis_functions)
eci_file = "data/ce_hydrostatic.json"
with open( eci_file, 'r' ) as infile:
ecis = json.load( infile )
print (ecis)
#calc = CE( ceBulk, ecis, size=(3,3,3) )
calc = get_ce_calc( ceBulk, kwargs, ecis, size=[10,10,10], free_unused_arrays_BC=True )
ceBulk = calc.BC
ceBulk.atoms.set_calculator( calc )
comp = {
"Al":1.0-mg_conc,
"Mg":mg_conc
}
calc.set_composition(comp)
mc_obj = Montecarlo( ceBulk.atoms, T, mpicomm=comm )
pairs = PairCorrelationObserver( calc )
network = NetworkObserver( calc=calc, cluster_name="c2_1414_1", element="Mg", nbins=100 )
mode = "fixed"
camera = Snapshot( "data/precipitation.traj", atoms=ceBulk.atoms )
mc_obj.attach( camera, interval=10000 )
mc_obj.attach( pairs, interval=1 )
mc_obj.attach( network, interval=500 )
mc_obj.runMC( mode=mode, steps=10000, equil=True )
pair_mean = pairs.get_average()
pair_std = pairs.get_std()
if ( rank == 0 ):
data = {
"pairs":pair_mean,
"pairs_std":pair_std,
"mg_conc":mg_conc,
"temperature":T
}
pairfname = "data/precipitation_pairs/precipitation_pairs_{}_{}K.json".format(int(1000*mg_conc),int(T))
with open(pairfname,'w') as outfile:
json.dump(data,outfile,indent=2, separators=(",",":"))
print ( "Thermal averaged pair correlation functions written to {}".format(pairfname) )
atoms = network.get_atoms_with_largest_cluster()
data = network.get_statistics() # This collects the histogram data from all processors
size,occurence = network.get_size_histogram()
data["histogram"] = {}
data["histogram"]["size"] = size.tolist()
data["histogram"]["occurence"] = occurence.tolist()
data["temperature"] = T
data["mg_conc"] = mg_conc
cluster_fname = "data/cluster_statistics_{}_{}K.json".format(int(1000*mg_conc),int(T))
atoms_fname = "data/largest_cluster_{}_{}K.cif".format( int(1000*mg_conc), int(T) )
if ( rank == 0 ):
print (occurence)
if ( rank == 0 ):
try:
with open( cluster_fname,'r') as infile:
data = json.load(infile)
old_size = np.array(data["histogram"]["size"])
old_hist = np.array(data["histogram"]["occurence"])
if ( np.allclose(old_size,size) ):
occurence += old_hist
data["histogram"]["occurence"] = occurence.tolist()
except Exception as exc:
print (str(exc))
with open( cluster_fname, 'w' ) as outfile:
json.dump( data, outfile, indent=2, separators=(",",":") )
print ("Cluster statistics written to {}".format(cluster_fname) )
write( atoms_fname, atoms )
print ("Atoms with largest cluster written to {}".format(atoms_fname))
#view(atoms)
#plt.plot( network.size_histogram, ls="steps")
#plt.show()
print ("Proc: {} reached final barrier".format(rank))
comm.barrier()
def thermodynamic_integration(phase):
alat = 3.8
conc_args = {}
conc_args['conc_ratio_min_1'] = [[1, 0]]
conc_args['conc_ratio_max_1'] = [[0, 1]]
kwargs = {
"crystalstructure": 'fcc',
"a": 3.8,
"size": [10, 10, 10],
"basis_elements": [['Cu', 'Au']],
"conc_args": conc_args,
"db_name": 'temp_sgc.db',
"max_cluster_size": 3,
"max_cluster_dist": 1.5 * alat
}
bc1 = BulkCrystal(**kwargs)
bf = bc1._get_basis_functions()[0]
with open("data/eci_aucu.json", 'r') as infile:
eci = json.load(infile)
db = dataset.connect("sqlite:///{}".format(canonical_db))
tbl = db["corrfunc"]
if phase == "Au" or phase == "Cu":
atoms1 = bc1.atoms
for atom in atoms1:
atom.symbol = phase
cf1 = get_pure_cf(eci, bf[phase])
else:
atoms = read(phase)
row = tbl.find_one(runID=gs[phase])
row.pop("id")
row.pop("runID")
cf1 = row
# TODO: Fix this when running with a pure phase
symbs = [atom.symbol for atom in atoms]
cf1 = get_pure_cf(eci, bf[bc1.atoms[0].symbol])
atoms = bc1.atoms
calc = CE(bc1, eci=eci, initial_cf=cf1)
calc.set_symbols(symbs)
atoms.set_calculator(calc)
sgc_db = dataset.connect("sqlite:///{}".format(sgc_db_name))
tbl = sgc_db["results"]
tbl_cf = sgc_db["corr_func"]
mu = 0.223
nsteps = 100 * len(atoms)
equil_param = {"mode": "fixed", "maxiter": 10 * len(atoms)}
order_param = SiteOrderParameter(atoms)
T = np.linspace(100, 600, 40)
# mu = np.linspace(0.223, 0.19, 20).tolist()
mu = [0.223]
for temp in T:
for m in mu:
chemical_potential = {"c1_0": m}
# mc = SGCMonteCarlo(atoms, temp, mpicomm=comm, symbols=["Au", "Cu"])
mc = Montecarlo(atoms, temp)
mc.attach(order_param)
init_formula = atoms.get_chemical_formula()
# mc.runMC(steps=nsteps, equil_params=equil_param,
# chem_potential=chemical_potential)
mc.runMC(steps=nsteps, equil_params=equil_param)
# thermo = mc.get_thermodynamic(reset_ecis=True)
thermo = mc.get_thermodynamic()
thermo["init_formula"] = init_formula
thermo["final_formula"] = atoms.get_chemical_formula()
avg, std = order_param.get_average()
thermo["order_avg"] = avg
thermo["order_std"] = std
thermo["valid"] = True
thermo["integration_path"] = "NN"
if rank == 0:
uid = tbl.insert(thermo)
cf = calc.get_cf()
cf["runID"] = uid
tbl_cf.insert(cf)
def run(mg_conc):
bs_kwargs = {
"conc_args": {
"conc_ratio_min_1": [[1, 0]],
"conc_ratio_max_1": [[0, 1]]
},
"basis_elements": [["Al", "Mg"], ["Al", "Mg"], ["Al", "Mg"],
["Al", "Mg"]],
"cellpar": [10.553, 10.553, 10.553, 90, 90, 90],
"basis": [(0, 0, 0), (0.324, 0.324, 0.324), (0.3582, 0.3582, 0.0393),
(0.0954, 0.0954, 0.2725)],
"spacegroup":
217,
"max_cluster_size":
4,
"db_name":
"trial_217.db",
"size": [1, 1, 1],
"grouped_basis": [[0, 1, 2, 3]]
}
bs = BulkSpacegroup(**bs_kwargs)
eci_file = "data/almg_217_eci.json"
with open(eci_file, 'r') as infile:
ecis = json.load(infile)
calc = get_ce_calc(bs, bs_kwargs, ecis, size=[3, 3, 3])
bs = calc.BC
bs.atoms.set_calculator(calc)
comp = {"Al": 1.0 - mg_conc, "Mg": mg_conc}
calc.set_composition(comp)
print("Number of atoms: {}".format(len(bs.atoms)))
high_temps = [10000, 9000, 8000, 7000, 6000, 5000, 4000, 3000, 2000, 1000]
low_temps = range(200, 1000, 50)[::-1]
T = np.array(high_temps + low_temps)
#T = np.array([10000,9000,8000,7000,6000,5000,4000,3000,2000,1000,800,700,600,500,400,375,350,325,300,275,250,225,200,175,150])
#T = np.array([1E6,100000])
precs = np.zeros(len(T)) + 1E-4
#precs[T<=500] = 1E-5
print(bs.atoms.get_chemical_formula())
mc_obj = Montecarlo(bs.atoms, T[0], mpicomm=comm)
mc_obj.accept_first_trial_move_after_reset = False
for prec, temp in zip(precs, T):
mc_obj.T = temp
mc_obj.reset()
#if ( temp==T[0] ):
# mc_obj.is_first = False # Do not accept the first move
print("Current temperature {}K".format(temp))
mode = "fixed"
mc_obj.runMC(mode=mode, prec=prec, steps=1000000)
thermo = mc_obj.get_thermodynamic()
thermo["temperature"] = temp
thermo["prec"] = prec
thermo["internal_energy"] = thermo.pop("energy")
thermo["converged"] = True
if (rank == 0):
cf = calc.get_cf()
db = connect(mc_db_name)
thermo.update(cf)
db.write(bs.atoms, key_value_pairs=thermo)
if (rank == 0 and run_mfa):
# At this point a ground state should have been reached
mfa = CanonicalMeanField(atoms=bs.atoms, T=T)
mfa.relax()
res = mfa.calculate()
formula = bs.atoms.get_chemical_formula()
fname = "data/mfa217/{}mfa.json".format(formula)
with open(fname, 'w') as outfile:
json.dump(res, outfile, indent=2, separators=(",", ":"))
def _accept(self, system_changes):
move_accepted = Montecarlo._accept(self, system_changes)
in_window = self.nuc_sampler.is_in_window(self.network)
return move_accepted and in_window
def free_energy_vs_layered(T, mod):
from cemc.mcmc import AdaptiveBiasReactionPathSampler
from cemc.mcmc import ReactionCrdRangeConstraint
from cemc.mcmc import DiffractionObserver
from cemc.mcmc import MinimalEnergyPath
from ase.geometry import get_layers
atoms = get_atoms(cubic=True)
atoms_cpy = atoms.copy()
layers, dist = get_layers(atoms, (0, 1, 0))
for atom in atoms_cpy:
if layers[atom.index] % 2 == 0:
atom.symbol = "Mg"
else:
atom.symbol = "Si"
symbols = [atom.symbol for atom in atoms_cpy]
atoms.get_calculator().set_symbols(symbols)
lamb = 4.05
k = 2.0 * np.pi / lamb
workdir = "data/diffraction"
# Have to perform only insert moves
mc = Montecarlo(atoms, T)
k_vec = [k, 0, 0]
observer = DiffractionObserver(atoms=atoms,
active_symbols=["Si"],
all_symbols=["Mg", "Si"],
k_vector=k_vec,
name="reflection")
if T < 250:
mc.max_attempts = 5
nsteps = 100 * len(atoms)
mep = MinimalEnergyPath(mc_obj=mc,
observer=observer,
value_name="reflection",
relax_steps=nsteps,
search_steps=nsteps,
traj_file="{}/mep_diffract{}K.traj".format(
workdir, T),
max_reac_crd=1999.0)
mep.run()
mep.save(fname="{}/mep_mep_diffract_path{}.csv".format(workdir, T))
else:
conc_cnst = ReactionCrdRangeConstraint(observer,
value_name="reflection")
conc_cnst.update_range([0, 0.5])
mc.add_constraint(conc_cnst)
reac_path = AdaptiveBiasReactionPathSampler(
mc_obj=mc,
react_crd=[0.0, 0.5],
observer=observer,
n_bins=500,
data_file="{}/layered_bias{}K.h5".format(workdir, T),
mod_factor=mod,
delete_db_if_exists=True,
mpicomm=None,
db_struct="{}/layered_bias_struct{}K.db".format(workdir, T),
react_crd_name="reflection",
ignore_equil_steps=False,
smear=2000)
reac_path.run()
reac_path.save()
