def main():
atoms = atoms_with_calc(50)
# mc = SoluteChainMC(atoms, 350, cluster_elements=["Mg", "Si"],
# cluster_names=["c2_01nn_0"])
T = [10]
snapshot = Snapshot(trajfile="data/solute_chain{}_nostrain.traj".format(T[0]), atoms=atoms)
first = True
nano_part = get_nanoparticle()
for temp in T:
print("Current temperature {}".format(T))
mc = FixedNucleusMC(
atoms, temp, network_name=["c2_01nn_0"],
network_element=["Mg", "Si"],
max_constraint_attempts=1E6)
backup = MCBackup(mc, overwrite_db_row=False, db_name="data/mc_solute_no_strain.db")
mc.attach(backup, interval=20000)
strain = get_strain_observer(mc)
mc.add_bias(strain)
if first:
first = False
#mc.grow_cluster({"Mg": 1000, "Si": 1000})
symbs = insert_nano_particle(atoms.copy(), nano_part)
mc.set_symbols(symbs)
tag_by_layer_type(mc.atoms)
cnst = ConstrainElementByTag(atoms=mc.atoms, element_by_tag=[["Mg", "Al"], ["Si", "Al"]])
mc.add_constraint(cnst)
#mc.build_chain({"Mg": 500, "Si": 500})
fix_layer = FixEdgeLayers(thickness=5.0, atoms=mc.atoms)
mc.add_constraint(fix_layer)
mc.attach(snapshot, interval=20000)
mc.runMC(steps=19000, init_cluster=False)
def run(N, use_bias):
#network_name=["c2_4p050_3", "c2_2p864_2"]
bc = init_bc(N)
mc = FixedNucleusMC(
bc.atoms, T, network_name=["c2_2p864_2"],
network_element=["Mg", "Si"], mpicomm=comm,
max_constraint_attempts=1E6)
#mc.max_allowed_constraint_pass_attempts = 1
nanop = get_nanoparticle()
symbs = insert_nano_particle(bc.atoms.copy(), nanop)
mc.set_symbols(symbs)
mc.init_cluster_info()
formula = "(I1+I2)/(2*I3)" # Needle
#formula = "2*I1/(I2+I3)" # Plate
inert_init = InertiaCrdInitializer(
fixed_nucl_mc=mc, matrix_element="Al", cluster_elements=["Mg", "Si"],
formula=formula)
inert_rng_constraint = InertiaRangeConstraint(
fixed_nuc_mc=mc, inertia_init=inert_init, verbose=True)
if use_bias:
bias = InertiaBiasPotential.load(fname=inertia_bias_file)
bias.inertia_range = inert_rng_constraint
mc.add_bias(bias)
nsteps = 100000
if rank == 0:
snap = Snapshot(atoms=bc.atoms, trajfile="{}/inertia.traj".format(workdir))
mc.attach(snap, interval=100000)
# reac_path = ReactionPathSampler(
# mc_obj=mc, react_crd=[0.05, 0.9], react_crd_init=inert_init,
# react_crd_range_constraint=inert_rng_constraint, n_windows=30,
# n_bins=10, data_file=h5file)
# reac_path.run(nsteps=nsteps)
# reac_path.save()
inert_rng_constraint.range = [0.0, 0.9]
fix_layer = FixEdgeLayers(thickness=5.0, atoms=mc.atoms)
mc.add_constraint(inert_rng_constraint)
mc.add_constraint(fix_layer)
reac_path = AdaptiveBiasReactionPathSampler(
mc_obj=mc, react_crd=[0.0, 0.9], react_crd_init=inert_init,
n_bins=100, data_file="{}/adaptive_bias.h5".format(workdir),
mod_factor=1E-5, delete_db_if_exists=True, mpicomm=None,
db_struct="{}/adaptive_bias_struct.db".format(workdir))
reac_path.run()
reac_path.save()
def test_covariance_observer(self):
"""Test the covariance observer."""
if not available:
self.skipTest("ASE version does not have CE!")
msg = ""
no_throw = True
from cemc.mcmc import FixEdgeLayers
from cemc.mcmc import CovarianceMatrixObserver
bc, args = get_ternary_BC(ret_args=True)
ecis = get_example_ecis(bc=bc)
atoms = get_atoms_with_ce_calc(bc, args, eci=ecis, size=[8, 8, 8], db_name="covariance_obs.db")
T = 200
nn_names = [name for name in bc.cluster_family_names
if int(name[1]) == 2]
mc = FixedNucleusMC(
atoms, T, network_name=nn_names,
network_element=["Mg", "Si"])
fixed_layers = FixEdgeLayers(atoms=mc.atoms, thickness=3.0)
mc.add_constraint(fixed_layers)
elements = {"Mg": 6, "Si": 6}
mc.insert_symbol_random_places("Mg", num=1, swap_symbs=["Al"])
mc.grow_cluster(elements)
cov_obs = CovarianceMatrixObserver(atoms=mc.atoms, cluster_elements=["Mg", "Si"])
mc.attach(cov_obs)
for _ in range(10):
mc.runMC(steps=100, elements=elements, init_cluster=False)
obs_I = cov_obs.cov_matrix
indices = []
for atom in mc.atoms:
if atom.symbol in ["Mg", "Si"]:
indices.append(atom.index)
cluster = mc.atoms[indices]
pos = cluster.get_positions()
com = np.mean(pos, axis=0)
pos -= com
cov_matrix = np.zeros((3, 3))
for i in range(pos.shape[0]):
x = pos[i, :]
cov_matrix += np.outer(x, x)
self.assertTrue(np.allclose(obs_I, cov_matrix))
os.remove("covariance_obs.db")