def test_read_write(self):
"""Tests read and write functionalities."""
# save to file
temp_file = tempfile.NamedTemporaryFile()
self.ce.write(temp_file.name)
# read from file
temp_file.seek(0)
ce_read = ClusterExpansion.read(temp_file.name)
# check cluster space
self.assertEqual(self.cs._input_structure,
ce_read._cluster_space._input_structure)
self.assertEqual(self.cs._cutoffs, ce_read._cluster_space._cutoffs)
self.assertEqual(self.cs._input_chemical_symbols,
ce_read._cluster_space._input_chemical_symbols)
# check parameters
self.assertIsInstance(ce_read.parameters, np.ndarray)
self.assertEqual(list(ce_read.parameters), list(self.parameters))
# check metadata
self.assertEqual(len(self.ce.metadata), len(ce_read.metadata))
self.assertSequenceEqual(sorted(self.ce.metadata.keys()),
sorted(ce_read.metadata.keys()))
for key in self.ce.metadata.keys():
self.assertEqual(self.ce.metadata[key], ce_read.metadata[key])
def test_read_write_pruned(self):
"""Tests read and write functionalities."""
# save to file
temp_file = tempfile.NamedTemporaryFile()
self.ce.prune(indices=[2, 3])
self.ce.prune(tol=3)
pruned_params = self.ce.parameters
pruned_cs_len = len(self.ce._cluster_space)
self.ce.write(temp_file.name)
# read from file
temp_file.seek(0)
ce_read = ClusterExpansion.read(temp_file.name)
params_read = ce_read.parameters
cs_len_read = len(ce_read._cluster_space)
# check cluster space
self.assertEqual(cs_len_read, pruned_cs_len)
self.assertEqual(list(params_read), list(pruned_params))
# This scripts runs in about 6 seconds on an i7-6700K CPU.
import matplotlib.pyplot as plt
from ase.db import connect
from icet import ClusterExpansion
# step 1: Compile predicted and reference data for plotting
ce = ClusterExpansion.read('mixing_energy.ce')
data = {'concentration': [], 'reference_energy': [], 'predicted_energy': []}
db = connect('reference_data.db')
for row in db.select('natoms<=6'):
data['concentration'].append(row.concentration)
# the factor of 1e3 serves to convert from eV/atom to meV/atom
data['reference_energy'].append(1e3 * row.mixing_energy)
data['predicted_energy'].append(1e3 * ce.predict(row.toatoms()))
# step 2: Plot results
fig, ax = plt.subplots(figsize=(4, 3))
ax.set_xlabel(r'Pd concentration')
ax.set_ylabel(r'Mixing energy (meV/atom)')
ax.set_xlim([0, 1])
ax.set_ylim([-69, 15])
ax.scatter(data['concentration'], data['reference_energy'],
marker='o', label='reference')
ax.scatter(data['concentration'], data['predicted_energy'],
marker='x', label='CE prediction')
plt.savefig('mixing_energy_comparison.png', bbox_inches='tight')
from ase.db import connect
from icet import (StructureContainer, CrossValidationEstimator,
ClusterExpansion)
from mchammer.ensembles import SemiGrandCanonicalEnsemble as SGCEnsemble
from mchammer.calculators import ClusterExpansionCalculator as CECalculator
from mchammer.observers import ClusterExpansionObserver as CEObserver
# step 1: Read cluster expansion for mixing energies from file
ce_mix_energies = ClusterExpansion.read('mixing_energy.ce')
chemical_symbols = ce_mix_energies.cluster_space.chemical_symbols[0]
cs = ce_mix_energies.cluster_space
# step 2: Read structures from database into a structure container
db = connect('reference_data.db')
sc = StructureContainer(cluster_space=cs)
for row in db.select('natoms<=8'):
sc.add_structure(structure=row.toatoms(),
user_tag=row.tag,
properties={'lattice_parameter': row.lattice_parameter})
# step 3: Construct cluster expansion for lattice parameter
fit_data = sc.get_fit_data(key='lattice_parameter')
opt = CrossValidationEstimator(fit_data=fit_data, fit_method='lasso')
opt.validate()
opt.train()
ce_latt_param = ClusterExpansion(cluster_space=cs, parameters=opt.parameters)
# step 4: Set up the calculator and a canonical ensemble
structure = cs.primitive_structure.repeat(3)
structure.set_chemical_symbols([chemical_symbols[0]] * len(structure))
calculator = CECalculator(structure=structure,