def __init__(self, *args, **kwargs):
kwargs['Hamiltonian_SCC'] = 'Yes' # SCC = self-consistency charges
kwargs['Hamiltonian_ShellResolvedSCC'] = 'No' # Use l-dependent Hubbard values ?
kwargs['Hamiltonian_OrbitalResolvedSCC'] = 'No'
kwargs['Hamiltonian_SCCTolerance'] = '1e-5' # SCC convergence criterion
kwargs['Hamiltonian_MaxSCCIterations'] = 250
kwargs['maximum_angular_momenta'] = {'Xx': MAM1, 'Yy': MAM2}
kwargs['Hamiltonian_Charge'] = '0.000000'
kwargs['Hamiltonian_ReadInitialCharges'] = 'No' # DFTB-equivalent of restarting from saved electron density
kwargs['Hamiltonian_Filling'] = 'Fermi {'
kwargs['Hamiltonian_Filling_empty'] = 'Temperature [Kelvin] = 300'
kwargs['Hamiltonian_PolynomialRepulsive'] = 'SetForAll {Yes}' # Use polynomial or spline repulsive ?
kwargs['Hamiltonian_Eigensolver'] = 'RelativelyRobust {}'
DftbPlusCalculator.__init__(self, *args, **kwargs)
def singlepoint(t):
''' This function will be used in the iter007 step, which restarts
from the runXXX/godb.db databases from the iter006 step.
Since the repulsive interactions have been reparametrized in
between, the structures in the database need to be re-evaluated.
'''
calc = DftbPlusCalculator(t, kpts=(1, 1, 1), use_spline=True,
maximum_angular_momenta={'Si': 1})
t.set_calculator(calc)
E = t.get_potential_energy()
F = t.get_forces()
S = None
finalize(t, energy=E, forces=F, stress=S)
return t
def singlepoint(t, kptdensity=3.5):
if get_raw_score(t) < -1e5:
return t
try:
calc = DftbPlusCalculator(t, kpts=kptdensity, use_spline=True,
maximum_angular_momenta={'Pd': 2, 'H': 0, 'O': 1})
t.set_calculator(calc)
E = t.get_potential_energy()
F = t.get_forces()
S = t.get_stress()
finalize(t, energy=E, forces=F, stress=S)
penalize(t)
except (IOError, TypeError, RuntimeError, UnboundLocalError) as err:
print(err)
print('Warning: problems with singlepoint recalculation')
finalize(t, energy=1e9, forces=None, stress=None)
return t
def relax_one(t):
''' This method defines how to locally minimize a given
atoms object 't'.
'''
# The provided structure will often contain atoms separated
# by relatively short distances. Pushing these atoms a bit
# apart using a soft potential will reduce the number of
# subsequent ionic steps and will help avoid DFTB convergence
# problems.
pos = t.get_positions()
numbers = list(set(t.get_atomic_numbers()))
blmin = closest_distances_generator(numbers, 0.5)
t = push_apart(t, blmin, variable_cell=False)
print('Starting relaxation', flush=True)
clock = time()
t.wrap()
# Define the DFTB+ calculator:
calc = DftbPlusCalculator(t, kpts=(1, 1, 1), use_spline=True,
maximum_angular_momenta={'Si': 1})
# Start the actual relaxation using the
# tango.relax_utils.relax_precon method.
# This wraps around the preconditioned
# optimizers in ASE and also takes care of
# setting the raw score etc.:
try:
t = relax_precon(t, calc, fmax=1e-2, variable_cell=False,
logfile=None, trajfile=None)
except IOError:
# the DFTB+ ASE calculator throws an IOError
# in case it couldn't find the 'results.tag'
# output file, which may sporadically happen
# due to SCC converge issues when e.g. a Si7
# cluster is breaking into fragments. We simply
# handle this by aborting the relaxation and
# assigning a very high energy to the structure:
finalize(t, energy=1e9, forces=None, stress=None)
print('Relaxing took %.3f seconds.' % (time() - clock), flush=True)
return t
def relax_one(t, kptdensity=3.5):
cellbounds = CellBounds(
bounds={
'phi': [0.1 * 180., 0.9 * 180.],
'chi': [0.1 * 180., 0.9 * 180.],
'psi': [0.1 * 180., 0.9 * 180.],
'a': [1.5, 20],
'b': [1.5, 20],
'c': [1.5, 20]
})
if not cellbounds.is_within_bounds(t.get_cell()):
print('Candidate outside cellbounds -- skipping')
finalize(t, energy=1e9, forces=None, stress=None)
return t
pos = t.get_positions()
numbers = list(set(t.get_atomic_numbers()))
blmin = closest_distances_generator(numbers, 0.5)
t = push_apart(t, blmin, variable_cell=True)
print('Starting relaxation', flush=True)
clock = time()
t.wrap()
calc = DftbPlusCalculator(t,
kpts=0.66 * kptdensity,
use_spline=True,
read_chg=True,
maximum_angular_momenta={'C': 1})
try:
t = relax_precon(t,
calc,
fmax=2e-1,
smax=1e-2,
variable_cell=True,
optimizer='LBFGS',
a_min=1e-4,
cellbounds=cellbounds,
logfile='opt_first.log',
trajfile='opt_first.traj')
except (IOError, TypeError, RuntimeError, UnboundLocalError) as err:
# SCC or geometry optimization convergence problem
print(err)
del t.constraints
t.wrap()
calc = DftbPlusCalculator(t,
kpts=kptdensity,
use_spline=True,
read_chg=True,
maximum_angular_momenta={'C': 1})
try:
t = relax_precon(t,
calc,
fmax=1e-1,
smax=5e-3,
variable_cell=True,
optimizer='LBFGS',
a_min=1e-4,
cellbounds=cellbounds,
logfile='opt.log',
trajfile='opt.traj')
except (IOError, TypeError, RuntimeError, UnboundLocalError) as err:
# SCC or geometry optimization convergence problem
print(err)
try:
t = read('opt.traj@-1')
energy = t.get_potential_energy()
forces = t.get_forces()
stress = t.get_stress()
except (FileNotFoundError, UnknownFileTypeError) as err:
print(err)
energy, forces, stress = (1e9, None, None)
finalize(t, energy=energy, forces=forces, stress=stress)
print('Relaxing took %.3f seconds.' % (time() - clock), flush=True)
os.system('mv opt_first.traj prev_first.traj')
os.system('mv opt_first.log prev_first.log')
os.system('mv opt.log prev.log')
os.system('mv opt.traj prev.traj')
penalize(t)
return t
def relax_one(t, kptdensity=3.5):
cellbounds = CellBounds(bounds={'phi': [0.1 * 180., 0.9 * 180.],
'chi': [0.1 * 180., 0.9 * 180.],
'psi': [0.1 * 180., 0.9 * 180.],
'a': [1.5, 20], 'b': [1.5, 20],
'c':[1.5, 20]})
if not cellbounds.is_within_bounds(t.get_cell()):
print('Candidate outside cellbounds -- skipping')
finalize(t, energy=1e9, forces=None, stress=None)
return t
tags = t.get_tags()
pos = t.get_positions()
pairs = []
for tag in list(set(tags)):
indices = list(np.where(tags == tag)[0])
if len(indices) == 2:
pairs.append(indices)
c = FixBondLengths(pairs)
t.set_constraint(c)
blmin = {(1, 1): 1.8, (1, 8): 0.9, (1, 46): 1.8, (8, 8): 2.0,
(8, 46): 1.5, (46, 46): 1.5}
t = push_apart(t, blmin, variable_cell=True)
del t.constraints
oh_bondlength = 0.97907
for (o_atom, h_atom) in pairs:
vec = t.get_distance(o_atom, h_atom, mic=True, vector=True)
pos[h_atom] = pos[o_atom] + vec * oh_bondlength / np.linalg.norm(vec)
t.set_positions(pos)
print('Starting relaxation', flush=True)
clock = time()
t.wrap()
calc = DftbPlusCalculator(t, kpts=0.66*kptdensity,
use_spline=True, read_chg=True,
maximum_angular_momenta={'Pd': 2, 'H': 0, 'O': 1})
try:
t = relax_precon(t, calc, fmax=2e-1, smax=1e-2, variable_cell=True,
optimizer='LBFGS', a_min=1e-4, cellbounds=cellbounds,
fix_bond_lengths_pairs=pairs, logfile='opt_first.log',
trajfile='opt_first.traj')
except (IOError, TypeError, RuntimeError, UnboundLocalError) as err:
# SCC or geometry optimization convergence problem
print(err)
if isinstance(t, Filter):
t = t.atoms
del t.constraints
t.wrap()
calc = DftbPlusCalculator(t, kpts=kptdensity,
use_spline=True, read_chg=True,
maximum_angular_momenta={'Pd': 2, 'H': 0, 'O': 1})
try:
t = relax_precon(t, calc, fmax=1e-1, smax=5e-3, variable_cell=True,
optimizer='LBFGS', a_min=1e-4, cellbounds=cellbounds,
fix_bond_lengths_pairs=pairs, logfile='opt.log',
trajfile='opt.traj')
except (IOError, TypeError, RuntimeError, UnboundLocalError) as err:
# SCC or geometry optimization convergence problem
print(err)
try:
t = read('opt.traj@-1')
energy = t.get_potential_energy()
forces = t.get_forces()
stress = t.get_stress()
except (FileNotFoundError, UnknownFileTypeError) as err:
print(err)
energy, forces, stress = (1e9, None, None)
if isinstance(t, Filter):
t = t.atoms
finalize(t, energy=energy, forces=forces, stress=stress)
print('Relaxing took %.3f seconds.' % (time() - clock), flush=True)
os.system('mv opt_first.traj prev_first.traj')
os.system('mv opt_first.log prev_first.log')
os.system('mv opt.log prev.log')
os.system('mv opt.traj prev.traj')
penalize(t)
return t
def __init__(self, *args, **kwargs):
kwargs['Hamiltonian_SCC'] = 'No'
kwargs['Hamiltonian_ShellResolvedSCC'] = 'No'
kwargs['Hamiltonian_OrbitalResolvedSCC'] = 'No'
kwargs['maximum_angular_momenta'] = {'Si': 1}
DftbPlusCalculator.__init__(self, *args, **kwargs)
