def test_gradients():
h2 = Molecule(name='h2', atoms=[Atom('H'), Atom('H', x=1.0)])
calc = Calculation(name='h2_grad',
molecule=h2,
method=method,
keywords=method.keywords.grad())
calc.run()
h2.energy = calc.get_energy()
delta_r = 1E-8
# Energy of a finite difference approximation
h2_disp = Molecule(name='h2_disp',
atoms=[Atom('H'), Atom('H', x=1.0 + delta_r)])
calc = Calculation(name='h2_disp',
molecule=h2_disp,
method=method,
keywords=method.keywords.grad)
calc.run()
h2_disp.energy = calc.get_energy()
delta_energy = h2_disp.energy - h2.energy # Ha
grad = delta_energy / delta_r # Ha A^-1
calc = Calculation(name='h2_grad',
molecule=h2,
method=method,
keywords=method.keywords.grad)
calc.run()
diff = calc.get_gradients()[1, 0] - grad # Ha A^-1
# Difference between the absolute and finite difference approximation
assert np.abs(diff) < 1E-3
def get_optimised_species(calc, method, direction, atoms):
"""Get the species that is optimised from an initial set of atoms"""
species = Molecule(name=f'{calc.name}_{direction}',
atoms=atoms,
charge=calc.molecule.charge,
mult=calc.molecule.mult)
# Note that for the surface to be the same the keywords.opt and keywords.hess need to match in the level of theory
calc = Calculation(name=f'{calc.name}_{direction}',
molecule=species,
method=method,
keywords=method.keywords.opt,
n_cores=Config.n_cores)
calc.run()
try:
species.set_atoms(atoms=calc.get_final_atoms())
species.energy = calc.get_energy()
make_graph(species)
except AtomsNotFound:
logger.error(f'{direction} displacement calculation failed')
return species
def test_grad():
h2 = Molecule(name='h2', atoms=[Atom('H'), Atom('H', x=0.5)])
grad_calc = Calculation(name='h2_grad',
molecule=h2,
method=method,
keywords=Config.MOPAC.keywords.grad)
grad_calc.run()
energy = grad_calc.get_energy()
assert energy is not None
gradients = grad_calc.get_gradients()
assert gradients.shape == (2, 3)
delta_r = 1E-5
h2_disp = Molecule(name='h2_disp',
atoms=[Atom('H'), Atom('H', x=0.5 + delta_r)])
h2_disp.single_point(method)
delta_energy = h2_disp.energy - energy # Ha]
grad = delta_energy / delta_r # Ha A^-1
# Difference between the absolute and finite difference approximation
assert np.abs(gradients[1, 0] - grad) < 1E-1
# Broken gradient file
grad_calc.output.filename = 'h2_grad_broken.out'
grad_calc.output.file_lines = open('h2_grad_broken.out', 'r').readlines()
with pytest.raises(CouldNotGetProperty):
_ = grad_calc.get_gradients()
def test_opt_calc():
calc = Calculation(name='opt',
molecule=test_mol,
method=method,
keywords=opt_keywords)
calc.run()
assert os.path.exists('opt_nwchem.nw')
assert os.path.exists('opt_nwchem.out')
final_atoms = calc.get_final_atoms()
assert len(final_atoms) == 5
assert type(final_atoms[0]) is Atom
assert -40.4165 < calc.get_energy() < -40.4164
assert calc.output.exists()
assert calc.output.file_lines is not None
assert calc.get_imaginary_freqs() == []
assert calc.input.filename == 'opt_nwchem.nw'
assert calc.output.filename == 'opt_nwchem.out'
assert calc.terminated_normally()
assert calc.optimisation_converged()
assert calc.optimisation_nearly_converged() is False
charges = calc.get_atomic_charges()
assert len(charges) == 5
assert all(-1.0 < c < 1.0 for c in charges)
# Optimisation should result in small gradients
gradients = calc.get_gradients()
assert len(gradients) == 5
assert all(-0.1 < np.linalg.norm(g) < 0.1 for g in gradients)
def test_mopac_opt_calculation():
calc = Calculation(name='opt',
molecule=methylchloride,
method=method,
keywords=Config.MOPAC.keywords.opt)
calc.run()
assert os.path.exists('opt_mopac.mop') is True
assert os.path.exists('opt_mopac.out') is True
assert len(calc.get_final_atoms()) == 5
# Actual energy in Hartrees
energy = Constants.eV2ha * -430.43191
assert energy - 0.0001 < calc.get_energy() < energy + 0.0001
assert calc.output.exists()
assert calc.output.file_lines is not None
assert calc.input.filename == 'opt_mopac.mop'
assert calc.output.filename == 'opt_mopac.out'
assert calc.terminated_normally()
assert calc.optimisation_converged() is True
with pytest.raises(CouldNotGetProperty):
_ = calc.get_gradients()
with pytest.raises(NotImplementedError):
_ = calc.optimisation_nearly_converged()
with pytest.raises(NotImplementedError):
_ = calc.get_imaginary_freqs()
with pytest.raises(NotImplementedError):
_ = calc.get_normal_mode_displacements(4)
def optimise(self, method, reset_graph=False, calc=None):
"""
Optimise the geometry of this conformer
Arguments:
method (autode.wrappers.base.ElectronicStructureMethod):
Keyword Arguments:
reset_graph (bool):
calc (autode.calculation.Calculation):
"""
logger.info(f'Running optimisation of {self.name}')
if calc is not None or reset_graph:
raise NotImplementedError
opt = Calculation(name=f'{self.name}_opt',
molecule=self,
method=method,
keywords=method.keywords.low_opt,
n_cores=Config.n_cores,
distance_constraints=self.dist_consts)
opt.run()
self.energy = opt.get_energy()
try:
self.set_atoms(atoms=opt.get_final_atoms())
except AtomsNotFound:
logger.error(f'Atoms not found for {self.name} but not critical')
self.set_atoms(atoms=None)
return None
def test_psi4_opt_calculation():
methylchloride = Molecule(name='CH3Cl',
smiles='[H]C([H])(Cl)[H]',
solvent_name='water')
calc = Calculation(name='opt',
molecule=methylchloride,
method=method,
keywords=opt_keywords)
calc.run()
assert os.path.exists('opt_psi4.inp') is True
assert os.path.exists('opt_orca.out') is True
assert len(calc.get_final_atoms()) == 5
assert -499.735 < calc.get_energy() < -499.730
assert calc.output.exists()
assert calc.output.file_lines is not None
assert calc.get_imaginary_freqs() == []
assert calc.input.filename == 'opt_psi4.inp'
assert calc.output.filename == 'opt_psi4.out'
assert calc.terminated_normally()
assert calc.optimisation_converged()
calc = Calculation(name='opt',
molecule=methylchloride,
method=method,
keywords=opt_keywords)
# If the calculation is not run with calc.run() then there should be no
# input and the calc should raise that there is no input
with pytest.raises(NoInputError):
execute_calc(calc)
def optimise(self, method=None, reset_graph=False, calc=None):
"""
Optimise the geometry using a method
Arguments:
method (autode.wrappers.base.ElectronicStructureMethod):
Keyword Arguments:
reset_graph (bool): Reset the molecular graph
calc (autode.calculation.Calculation): Different e.g. constrained
optimisation calculation
"""
logger.info(f'Running optimisation of {self.name}')
if calc is None:
assert method is not None
calc = Calculation(name=f'{self.name}_opt', molecule=self,
method=method, keywords=method.keywords.opt,
n_cores=Config.n_cores)
else:
assert isinstance(calc, Calculation)
calc.run()
self.energy = calc.get_energy()
self.set_atoms(atoms=calc.get_final_atoms())
self.print_xyz_file(filename=f'{self.name}_optimised_{method.name}.xyz')
if reset_graph:
make_graph(self)
return None
def test_other_spin_states():
o_singlet = Molecule(atoms=[Atom('O')], mult=1)
o_singlet.name = 'molecule'
calc = Calculation(name='O_singlet',
molecule=o_singlet,
method=method,
keywords=Config.MOPAC.keywords.sp)
calc.run()
singlet_energy = calc.get_energy()
o_triplet = Molecule(atoms=[Atom('O')], mult=3)
o_triplet.name = 'molecule'
calc = Calculation(name='O_triplet',
molecule=o_triplet,
method=method,
keywords=Config.MOPAC.keywords.sp)
calc.run()
triplet_energy = calc.get_energy()
assert triplet_energy < singlet_energy
h_doublet = Molecule(atoms=[Atom('H')], mult=2)
h_doublet.name = 'molecule'
calc = Calculation(name='h',
molecule=h_doublet,
method=method,
keywords=Config.MOPAC.keywords.sp)
calc.run()
# Open shell doublet should work
assert calc.get_energy() is not None
h_quin = Molecule(atoms=[Atom('H')], mult=5)
h_quin.name = 'molecule'
with pytest.raises(UnsuppportedCalculationInput):
calc = Calculation(name='h',
molecule=h_quin,
method=method,
keywords=Config.MOPAC.keywords.sp)
calc.run()
os.chdir(here)
def singlepoint(molecule, method, keywords, n_cores=None):
"""
Run a single point energy evaluation on a molecule
:param molecule: (object)
:param method: (autode.ElectronicStructureMethod)
:param keywords: (list(str)) Keywords to use for the electronic structure
calculation e.g. ['Opt', 'PBE', 'def2-SVP']
:param n_cores: (int) Number of cores to use
:return:
"""
logger.info('Running single point calculation')
n_cores = Config.n_cores if n_cores is None else int(n_cores)
try:
from autode.calculation import Calculation
from autode.wrappers.XTB import xtb
from autode.wrappers.ORCA import orca
from autode.wrappers.keywords import SinglePointKeywords
except ModuleNotFoundError:
logger.error('autode not found. Calculations not available')
raise RequiresAutodE
if keywords is None:
if method == orca:
keywords = SinglePointKeywords(
['SP', 'M062X', 'def2-TZVP', 'RIJCOSX', 'def2/J', 'SlowConv'])
logger.warning('No keywords were set for the single point but an '
'ORCA calculation was requested. '
f'Using {str(keywords)}')
elif method == xtb:
keywords = xtb.keywords.sp
else:
logger.critical('No keywords were set for the single-point '
'calculation')
raise Exception
else:
# If the keywords are specified as a list convert them to a set of
# OptKeywords, required for autodE
if type(keywords) is list:
keywords = SinglePointKeywords(keywords)
sp = Calculation(name=molecule.name + '_sp',
molecule=molecule,
method=method,
keywords=keywords,
n_cores=n_cores)
sp.run()
molecule.energy = sp.get_energy()
return None
def test_constrained_opt():
methane = Molecule(name='methane', smiles='C')
calc = Calculation(name='methane_opt', molecule=methane,
method=method,
keywords=Config.MOPAC.keywords.opt)
calc.run()
opt_energy = calc.get_energy()
# Constrained optimisation with a C–H distance of 1.2 Å
# (carbon is the first atom in the file)
const = Calculation(name='methane_const', molecule=methane,
method=method,
keywords=Config.MOPAC.keywords.opt,
distance_constraints={(0, 1): 1.2})
const.run()
assert opt_energy < const.get_energy()
def single_point(self, method):
"""Calculate the single point energy of the species with a
autode.wrappers.base.ElectronicStructureMethod"""
logger.info(f'Running single point energy evaluation of {self.name}')
sp = Calculation(name=f'{self.name}_sp', molecule=self, method=method,
keywords=method.keywords.sp, n_cores=Config.n_cores)
sp.run()
self.energy = sp.get_energy()
return None
def test_bad_gauss_output():
calc = Calculation(name='no_output', molecule=test_mol, method=method,
keywords=opt_keywords)
calc.output_file_lines = []
calc.rev_output_file_lines = []
assert calc.get_energy() is None
with pytest.raises(AtomsNotFound):
calc.get_final_atoms()
with pytest.raises(NoInputError):
calc.execute_calculation()
def test_bad_geometry():
# Calculation with the wrong spin state should fail
calc = Calculation(name='h2_overlap_opt',
molecule=Molecule(atoms=[Atom('H'), Atom('H')]),
method=method,
keywords=Config.MOPAC.keywords.opt)
calc.output.filename = 'h2_overlap_opt_mopac.out'
calc.output.file_lines = open(calc.output.filename, 'r').readlines()
assert not calc.terminated_normally()
assert calc.get_energy() is None
assert not calc.optimisation_converged()
def test_point_charge_calc():
os.chdir(os.path.join(here, 'data'))
# Methane single point using a point charge with a unit positive charge
# located at (10, 10, 10)
calc = Calculation(
name='methane_point_charge',
molecule=test_mol,
method=method,
keywords=sp_keywords,
point_charges=[PointCharge(charge=1.0, x=10.0, y=10.0, z=10.0)])
calc.run()
# Assert that the input file is in the expected configuration
for line in open('methane_point_charge_g09.com', 'r'):
if 'PBE' in line:
assert 'Charge' in line
if len(line.split()) == 4:
if not line.split()[0].isdigit():
continue
x, y, z, charge = line.split()
assert float(x) == 10.0
assert float(y) == 10.0
assert float(z) == 10.0
assert float(charge) == 1.0
assert -40.428 < calc.get_energy() < -40.427
# Gaussian needs x-matrix and nosymm in the input line to run optimisations
# with point charges..
for opt_keyword in ['Opt', 'Opt=Tight', 'Opt=(Tight)']:
calc = Calculation(
name='methane_point_charge_o',
molecule=test_mol,
method=method,
keywords=OptKeywords(['PBE1PBE/Def2SVP', opt_keyword]),
point_charges=[PointCharge(charge=1.0, x=3.0, y=3.0, z=3.0)])
calc.generate_input()
for line in open('methane_point_charge_o_g09.com', 'r').readlines():
if 'PBE' in line:
assert 'charge' in line.lower()
assert 'z-matrix' in line.lower() and 'nosymm' in line.lower()
break
os.remove('methane_point_charge_g09.com')
os.remove('methane_point_charge_o_g09.com')
os.chdir(os.path.join(here))
def test_xtb_calculation():
test_mol = Molecule(name='test_mol',
smiles='O=C(C=C1)[C@@](C2NC3C=C2)([H])[C@@]3([H])C1=O')
calc = Calculation(name='opt', molecule=test_mol, method=method,
keywords=Config.XTB.keywords.opt)
calc.run()
assert os.path.exists('opt_xtb.xyz') is True
assert os.path.exists('opt_xtb.out') is True
assert len(calc.get_final_atoms()) == 22
assert calc.get_energy() == -36.990267613593
assert calc.output.exists()
assert calc.output.file_lines is not None
assert calc.input.filename == 'opt_xtb.xyz'
assert calc.output.filename == 'opt_xtb.out'
with pytest.raises(NotImplementedError):
calc.optimisation_nearly_converged()
with pytest.raises(NotImplementedError):
calc.get_imaginary_freqs()
with pytest.raises(NotImplementedError):
calc.get_normal_mode_displacements(4)
charges = calc.get_atomic_charges()
assert len(charges) == 22
assert all(-1.0 < c < 1.0 for c in charges)
const_opt = Calculation(name='const_opt', molecule=test_mol,
method=method,
distance_constraints={(0, 1): 1.2539792},
cartesian_constraints=[0],
keywords=Config.XTB.keywords.opt)
const_opt.generate_input()
assert os.path.exists('const_opt_xtb.xyz')
assert os.path.exists('xcontrol_const_opt_xtb')
const_opt.clean_up(force=True)
assert not os.path.exists('xcontrol_const_opt_xtb')
# Write an empty output file
open('tmp.out', 'w').close()
const_opt.output.filename = 'tmp.out'
const_opt.output.set_lines()
# cannot get atoms from an empty file
with pytest.raises(AtomsNotFound):
_ = const_opt.get_final_atoms()
def test_point_charge():
os.chdir(os.path.join(here, 'data', 'xtb'))
test_mol = Molecule(name='test_mol', smiles='C')
# Methane with a point charge fairly far away
calc = Calculation(name='opt_point_charge',
molecule=test_mol,
method=method,
keywords=Config.XTB.keywords.opt,
point_charges=[PointCharge(charge=1.0, x=10, y=1, z=1)])
calc.run()
assert -4.178 < calc.get_energy() < -4.175
os.chdir(here)
def test_mopac_with_pc():
calc = Calculation(name='opt_pc', molecule=methylchloride,
method=method,
keywords=Config.MOPAC.keywords.opt,
point_charges=[PointCharge(1, x=4, y=4, z=4)])
calc.run()
assert os.path.exists('opt_pc_mopac.mop') is True
assert os.path.exists('opt_pc_mopac.out') is True
assert len(calc.get_final_atoms()) == 5
# Actual energy in Hartrees without any point charges
energy = Constants.eV2ha * -430.43191
assert np.abs(calc.get_energy() - energy) > 0.0001
def run_autode(configuration, max_force=None, method=None, n_cores=1):
"""
Run an orca or xtb calculation
--------------------------------------------------------------------------
:param configuration: (gaptrain.configurations.Configuration)
:param max_force: (float) or None
:param method: (autode.wrappers.base.ElectronicStructureMethod)
"""
from autode.species import Species
from autode.calculation import Calculation
from autode.exceptions import CouldNotGetProperty
if method.name == 'orca' and GTConfig.orca_keywords is None:
raise ValueError("For ORCA training GTConfig.orca_keywords must be"
" set. e.g. "
"GradientKeywords(['PBE', 'def2-SVP', 'EnGrad'])")
# optimisation is not implemented, needs a method to run
assert max_force is None and method is not None
species = Species(name=configuration.name,
atoms=configuration.atoms,
charge=configuration.charge,
mult=configuration.mult)
# allow for an ORCA calculation to have non-default keywords.. not the
# cleanest implementation..
kwds = GTConfig.orca_keywords if method.name == 'orca' else method.keywords.grad
calc = Calculation(name='tmp',
molecule=species,
method=method,
keywords=kwds,
n_cores=n_cores)
calc.run()
ha_to_ev = 27.2114
try:
configuration.forces = -ha_to_ev * calc.get_gradients()
except CouldNotGetProperty:
logger.error('Failed to set forces')
configuration.energy = ha_to_ev * calc.get_energy()
configuration.partial_charges = calc.get_atomic_charges()
return configuration
def test_orca_opt_calculation():
os.chdir(os.path.join(here, 'data'))
methylchloride = Molecule(name='CH3Cl',
smiles='[H]C([H])(Cl)[H]',
solvent_name='water')
calc = Calculation(name='opt', molecule=methylchloride, method=method,
keywords=opt_keywords)
calc.run()
assert os.path.exists('opt_orca.inp') is True
assert os.path.exists('opt_orca.out') is True
assert len(calc.get_final_atoms()) == 5
assert -499.735 < calc.get_energy() < -499.730
assert calc.output.exists()
assert calc.output.file_lines is not None
assert calc.get_imaginary_freqs() == []
assert calc.input.filename == 'opt_orca.inp'
assert calc.output.filename == 'opt_orca.out'
assert calc.terminated_normally()
assert calc.optimisation_converged()
assert calc.optimisation_nearly_converged() is False
with pytest.raises(NoNormalModesFound):
calc.get_normal_mode_displacements(mode_number=0)
# Should have a partial atomic charge for every atom
charges = calc.get_atomic_charges()
assert len(charges) == 5
assert type(charges[0]) == float
assert -1.0 < charges[0] < 1.0
calc = Calculation(name='opt', molecule=methylchloride, method=method,
keywords=opt_keywords)
# If the calculation is not run with calc.run() then there should be no
# input and the calc should raise that there is no input
with pytest.raises(NoInputError):
execute_calc(calc)
os.remove('opt_orca.inp')
os.chdir(here)
def test_gauss_opt_calc():
os.chdir(os.path.join(here, 'data'))
methylchloride = Molecule(name='CH3Cl',
smiles='[H]C([H])(Cl)[H]',
solvent_name='water')
calc = Calculation(name='opt',
molecule=methylchloride,
method=method,
keywords=opt_keywords)
calc.run()
assert os.path.exists('opt_g09.com')
assert os.path.exists('opt_g09.log')
assert len(calc.get_final_atoms()) == 5
assert os.path.exists('opt_g09.xyz')
assert calc.get_energy() == -499.729222331
assert calc.output.exists()
assert calc.output.file_lines is not None
assert calc.get_imaginary_freqs() == []
with pytest.raises(NoNormalModesFound):
calc.get_normal_mode_displacements(mode_number=1)
assert calc.input.filename == 'opt_g09.com'
assert calc.output.filename == 'opt_g09.log'
assert calc.terminated_normally()
assert calc.optimisation_converged()
assert calc.optimisation_nearly_converged() is False
charges = calc.get_atomic_charges()
assert len(charges) == methylchloride.n_atoms
# Should be no very large atomic charges in this molecule
assert all(-1.0 < c < 1.0 for c in charges)
gradients = calc.get_gradients()
assert len(gradients) == methylchloride.n_atoms
assert len(gradients[0]) == 3
# Should be no large forces for an optimised molecule
assert sum(gradients[0]) < 0.1
os.remove('opt_g09.com')
os.chdir(here)
def test_xtb_calculation():
os.chdir(os.path.join(here, 'data'))
XTB.available = True
test_mol = Molecule(name='test_mol',
smiles='O=C(C=C1)[C@@](C2NC3C=C2)([H])[C@@]3([H])C1=O')
calc = Calculation(name='opt', molecule=test_mol, method=method,
keywords=Config.XTB.keywords.opt)
calc.run()
assert os.path.exists('opt_xtb.xyz') is True
assert os.path.exists('opt_xtb.out') is True
assert len(calc.get_final_atoms()) == 22
assert calc.get_energy() == -36.990267613593
assert calc.output.exists()
assert calc.output.file_lines is not None
assert calc.input.filename == 'opt_xtb.xyz'
assert calc.output.filename == 'opt_xtb.out'
with pytest.raises(NotImplementedError):
calc.optimisation_nearly_converged()
with pytest.raises(NotImplementedError):
calc.get_imaginary_freqs()
with pytest.raises(NotImplementedError):
calc.get_normal_mode_displacements(4)
charges = calc.get_atomic_charges()
assert len(charges) == 22
assert all(-1.0 < c < 1.0 for c in charges)
const_opt = Calculation(name='const_opt', molecule=test_mol,
method=method,
distance_constraints={(0, 1): 1.2539792},
cartesian_constraints=[0],
keywords=Config.XTB.keywords.opt)
const_opt.generate_input()
assert os.path.exists('xcontrol_const_opt_xtb')
os.remove('const_opt_xtb.xyz')
os.remove('xcontrol_const_opt_xtb')
os.remove('opt_xtb.xyz')
os.chdir(here)
class TransitionState(TSbase):
@requires_graph()
def _update_graph(self):
"""Update the molecular graph to include all the bonds that are being
made/broken"""
if self.bond_rearrangement is None:
logger.warning('Bond rearrangement not set - molecular graph '
'updating with no active bonds')
active_bonds = []
else:
active_bonds = self.bond_rearrangement.all
set_active_mol_graph(species=self, active_bonds=active_bonds)
logger.info(f'Molecular graph updated with active bonds')
return None
def _run_opt_ts_calc(self, method, name_ext):
"""Run an optts calculation and attempt to set the geometry, energy and
normal modes"""
if self.bond_rearrangement is None:
logger.warning('Cannot add redundant internal coordinates for the '
'active bonds with no bond rearrangement')
bond_ids = None
else:
bond_ids = self.bond_rearrangement.all
self.optts_calc = Calculation(
name=f'{self.name}_{name_ext}',
molecule=self,
method=method,
n_cores=Config.n_cores,
keywords=method.keywords.opt_ts,
bond_ids_to_add=bond_ids,
other_input_block=method.keywords.optts_block)
self.optts_calc.run()
if not self.optts_calc.optimisation_converged():
if self.optts_calc.optimisation_nearly_converged():
logger.info('Optimisation nearly converged')
self.calc = self.optts_calc
if self.could_have_correct_imag_mode():
logger.info('Still have correct imaginary mode, trying '
'more optimisation steps')
self.atoms = self.optts_calc.get_final_atoms()
self.optts_calc = Calculation(
name=f'{self.name}_{name_ext}_reopt',
molecule=self,
method=method,
n_cores=Config.n_cores,
keywords=method.keywords.opt_ts,
bond_ids_to_add=bond_ids,
other_input_block=method.keywords.optts_block)
self.optts_calc.run()
else:
logger.info('Lost imaginary mode')
else:
logger.info('Optimisation did not converge')
try:
self.imaginary_frequencies = self.optts_calc.get_imaginary_freqs()
self.atoms = self.optts_calc.get_final_atoms()
self.energy = self.optts_calc.get_energy()
except (AtomsNotFound, NoNormalModesFound):
logger.error('Transition state optimisation calculation failed')
return
def _generate_conformers(self, n_confs=None):
"""Generate conformers at the TS """
from autode.conformers.conformer import Conformer
from autode.conformers.conf_gen import get_simanl_atoms
from autode.conformers.conformers import conf_is_unique_rmsd
n_confs = Config.num_conformers if n_confs is None else n_confs
self.conformers = []
distance_consts = get_distance_constraints(self)
with Pool(processes=Config.n_cores) as pool:
results = [
pool.apply_async(get_simanl_atoms, (self, distance_consts, i))
for i in range(n_confs)
]
conf_atoms_list = [res.get(timeout=None) for res in results]
for i, atoms in enumerate(conf_atoms_list):
conf = Conformer(name=f'{self.name}_conf{i}',
charge=self.charge,
mult=self.mult,
atoms=atoms,
dist_consts=distance_consts)
# If the conformer is unique on an RMSD threshold
if conf_is_unique_rmsd(conf, self.conformers):
conf.solvent = self.solvent
conf.graph = deepcopy(self.graph)
self.conformers.append(conf)
logger.info(f'Generated {len(self.conformers)} conformer(s)')
return None
@requires_atoms()
def print_imag_vector(self, mode_number=6, name=None):
"""Print a .xyz file with multiple structures visualising the largest
magnitude imaginary mode
Keyword Arguments:
mode_number (int): Number of the normal mode to visualise,
6 (default) is the lowest frequency vibration
i.e. largest magnitude imaginary, if present
name (str):
"""
assert self.optts_calc is not None
name = self.name if name is None else name
disp = -0.5
for i in range(40):
atoms = get_displaced_atoms_along_mode(
calc=self.optts_calc,
mode_number=int(mode_number),
disp_magnitude=disp,
atoms=self.atoms)
atoms_to_xyz_file(atoms=atoms, filename=f'{name}.xyz', append=True)
# Add displacement so the final set of atoms are +0.5 Å displaced
# along the mode, then displaced back again
sign = 1 if i < 20 else -1
disp += sign * 1.0 / 20.0
return None
@requires_atoms()
def optimise(self, name_ext='optts'):
"""Optimise this TS to a true TS """
logger.info(f'Optimising {self.name} to a transition state')
self._run_opt_ts_calc(method=get_hmethod(), name_ext=name_ext)
# A transition state is a first order saddle point i.e. has a single
# imaginary frequency
if len(self.imaginary_frequencies) == 1:
logger.info('Found a TS with a single imaginary frequency')
return
if len(self.imaginary_frequencies) == 0:
logger.error('Transition state optimisation did not return any '
'imaginary frequencies')
return
if all([freq > -50 for freq in self.imaginary_frequencies[1:]]):
logger.warning('Had small imaginary modes - not displacing along')
return
# There is more than one imaginary frequency. Will assume that the most
# negative is the correct mode..
for disp_magnitude in [1, -1]:
logger.info('Displacing along second imaginary mode to try and '
'remove')
dis_name_ext = name_ext + '_dis' if disp_magnitude == 1 else name_ext + '_dis2'
atoms, energy, calc = deepcopy(self.atoms), deepcopy(
self.energy), deepcopy(self.optts_calc)
self.atoms = get_displaced_atoms_along_mode(
self.optts_calc, mode_number=7, disp_magnitude=disp_magnitude)
self._run_opt_ts_calc(method=get_hmethod(), name_ext=dis_name_ext)
if len(self.imaginary_frequencies) == 1:
logger.info('Displacement along second imaginary mode '
'successful. Now have 1 imaginary mode')
break
self.optts_calc = calc
self.atoms = atoms
self.energy = energy
self.imaginary_frequencies = self.optts_calc.get_imaginary_freqs()
return None
def calc_g_cont(self, method=None, calc=None, temp=None):
"""Calculate the free energy (G) contribution"""
return super().calc_g_cont(method=method,
calc=self.optts_calc,
temp=temp)
def calc_h_cont(self, method=None, calc=None, temp=None):
"""Calculate the enthalpy (H) contribution"""
return super().calc_h_cont(method=method,
calc=self.optts_calc,
temp=temp)
def find_lowest_energy_ts_conformer(self, rmsd_threshold=None):
"""Find the lowest energy transition state conformer by performing
constrained optimisations"""
logger.info('Finding lowest energy TS conformer')
atoms, energy = deepcopy(self.atoms), deepcopy(self.energy)
calc = deepcopy(self.optts_calc)
hmethod = get_hmethod() if Config.hmethod_conformers else None
self.find_lowest_energy_conformer(hmethod=hmethod)
# Remove similar TS conformer that are similar to this TS based on root
# mean squared differences in their structures
thresh = Config.rmsd_threshold if rmsd_threshold is None else rmsd_threshold
self.conformers = [
conf for conf in self.conformers
if calc_heavy_atom_rmsd(conf.atoms, atoms) > thresh
]
logger.info(f'Generated {len(self.conformers)} unique (RMSD > '
f'{thresh} Å) TS conformer(s)')
# Optimise the lowest energy conformer to a transition state - will
# .find_lowest_energy_conformer will have updated self.atoms etc.
if len(self.conformers) > 0:
self.optimise(name_ext='optts_conf')
if self.is_true_ts() and self.energy < energy:
logger.info('Conformer search successful')
return None
# Ensure the energy has a numerical value, so a difference can be
# evaluated
self.energy = self.energy if self.energy is not None else 0
logger.warning(f'Transition state conformer search failed '
f'(∆E = {energy - self.energy:.4f} Ha). Reverting')
logger.info('Reverting to previously found TS')
self.atoms = atoms
self.energy = energy
self.optts_calc = calc
self.imaginary_frequencies = calc.get_imaginary_freqs()
return None
def is_true_ts(self):
"""Is this TS a 'true' TS i.e. has at least on imaginary mode in the
hessian and is the correct mode"""
if self.energy is None:
logger.warning('Cannot be true TS with no energy')
return False
if len(self.imaginary_frequencies) > 0:
if self.has_correct_imag_mode(calc=self.optts_calc):
logger.info('Found a transition state with the correct '
'imaginary mode & links reactants and products')
return True
return False
def save_ts_template(self, folder_path=None):
"""Save a transition state template containing the active bond lengths,
solvent and charge in folder_path
Keyword Arguments:
folder_path (str): folder to save the TS template to
(default: {None})
"""
if self.bond_rearrangement is None:
raise ValueError('Cannot save a TS template without a bond '
'rearrangement')
logger.info(f'Saving TS template for {self.name}')
truncated_graph = get_truncated_active_mol_graph(self.graph)
for bond in self.bond_rearrangement.all:
truncated_graph.edges[bond]['distance'] = self.distance(*bond)
ts_template = TStemplate(truncated_graph, species=self)
ts_template.save(folder_path=folder_path)
logger.info('Saved TS template')
return None
def __init__(self, ts_guess):
"""
Transition State
Arguments:
ts_guess (autode.transition_states.ts_guess.TSguess):
"""
super().__init__(atoms=ts_guess.atoms,
reactant=ts_guess.reactant,
product=ts_guess.product,
name=f'TS_{ts_guess.name}',
charge=ts_guess.charge,
mult=ts_guess.mult)
self.bond_rearrangement = ts_guess.bond_rearrangement
self.conformers = None
self.optts_calc = None
self.imaginary_frequencies = []
self._update_graph()
def test_calc_class():
xtb = XTB()
calc = Calculation(name='-tmp',
molecule=test_mol,
method=xtb,
keywords=xtb.keywords.sp)
# Should prepend a dash to appease some EST methods
assert not calc.name.startswith('-')
assert calc.molecule is not None
assert calc.method.name == 'xtb'
assert calc.get_energy() is None
assert calc.get_enthalpy() is None
assert calc.get_free_energy() is None
assert not calc.optimisation_converged()
assert not calc.optimisation_nearly_converged()
with pytest.raises(ex.AtomsNotFound):
_ = calc.get_final_atoms()
with pytest.raises(ex.CouldNotGetProperty):
_ = calc.get_gradients()
with pytest.raises(ex.CouldNotGetProperty):
_ = calc.get_atomic_charges()
# Calculation that has not been run shouldn't have an opt converged
assert not calc.optimisation_converged()
assert not calc.optimisation_nearly_converged()
# With a filename that doesn't exist a NoOutput exception should be raised
calc.output.filename = '/a/path/that/does/not/exist/tmp'
with pytest.raises(ex.NoCalculationOutput):
calc.output.set_lines()
# With no output should not be able to get properties
calc.output.filename = 'tmp'
calc.output.file_lines = []
with pytest.raises(ex.CouldNotGetProperty):
_ = calc.get_atomic_charges()
# or final atoms
with pytest.raises(ex.AtomsNotFound):
_ = calc.get_final_atoms()
# Should default to a single core
assert calc.n_cores == 1
calc_str = str(calc)
new_calc = Calculation(name='tmp2',
molecule=test_mol,
method=xtb,
keywords=xtb.keywords.sp)
new_calc_str = str(new_calc)
# Calculation strings need to be unique
assert new_calc_str != calc_str
new_calc = Calculation(name='tmp2',
molecule=test_mol,
method=xtb,
keywords=xtb.keywords.sp,
temp=5000)
assert str(new_calc) != new_calc_str
mol_no_atoms = Molecule()
with pytest.raises(ex.NoInputError):
_ = Calculation(name='tmp2',
molecule=mol_no_atoms,
method=xtb,
keywords=xtb.keywords.sp)
class TransitionState(TSbase):
@requires_graph()
def _update_graph(self):
"""Update the molecular graph to include all the bonds that are being
made/broken"""
set_active_mol_graph(species=self, active_bonds=self.bond_rearrangement.all)
logger.info(f'Molecular graph updated with '
f'{len(self.bond_rearrangement.all)} active bonds')
return None
def _run_opt_ts_calc(self, method, name_ext):
"""Run an optts calculation and attempt to set the geometry, energy and
normal modes"""
self.optts_calc = Calculation(name=f'{self.name}_{name_ext}',
molecule=self,
method=method,
n_cores=Config.n_cores,
keywords=method.keywords.opt_ts,
bond_ids_to_add=self.bond_rearrangement.all,
other_input_block=method.keywords.optts_block)
self.optts_calc.run()
if not self.optts_calc.optimisation_converged():
if self.optts_calc.optimisation_nearly_converged():
logger.info('Optimisation nearly converged')
self.calc = self.optts_calc
if self.could_have_correct_imag_mode():
logger.info('Still have correct imaginary mode, trying '
'more optimisation steps')
self.set_atoms(atoms=self.optts_calc.get_final_atoms())
self.optts_calc = Calculation(name=f'{self.name}_{name_ext}_reopt',
molecule=self,
method=method,
n_cores=Config.n_cores,
keywords=method.keywords.opt_ts,
bond_ids_to_add=self.bond_rearrangement.all,
other_input_block=method.keywords.optts_block)
self.optts_calc.run()
else:
logger.info('Lost imaginary mode')
else:
logger.info('Optimisation did not converge')
try:
self.imaginary_frequencies = self.optts_calc.get_imaginary_freqs()
self.set_atoms(atoms=self.optts_calc.get_final_atoms())
self.energy = self.optts_calc.get_energy()
except (AtomsNotFound, NoNormalModesFound):
logger.error('Transition state optimisation calculation failed')
return
def _generate_conformers(self, n_confs=None):
"""Generate conformers at the TS """
from autode.conformers.conformer import Conformer
from autode.conformers.conf_gen import get_simanl_atoms
from autode.conformers.conformers import conf_is_unique_rmsd
n_confs = Config.num_conformers if n_confs is None else n_confs
self.conformers = []
distance_consts = get_distance_constraints(self)
with Pool(processes=Config.n_cores) as pool:
results = [pool.apply_async(get_simanl_atoms, (self, distance_consts, i)) for i in range(n_confs)]
conf_atoms_list = [res.get(timeout=None) for res in results]
for i, atoms in enumerate(conf_atoms_list):
conf = Conformer(name=f'{self.name}_conf{i}', charge=self.charge,
mult=self.mult, atoms=atoms,
dist_consts=distance_consts)
# If the conformer is unique on an RMSD threshold
if conf_is_unique_rmsd(conf, self.conformers):
conf.solvent = self.solvent
conf.graph = deepcopy(self.graph)
self.conformers.append(conf)
logger.info(f'Generated {len(self.conformers)} conformer(s)')
return None
@requires_atoms()
def optimise(self, name_ext='optts'):
"""Optimise this TS to a true TS """
logger.info(f'Optimising {self.name} to a transition state')
self._run_opt_ts_calc(method=get_hmethod(), name_ext=name_ext)
# A transition state is a first order saddle point i.e. has a single
# imaginary frequency
if len(self.imaginary_frequencies) == 1:
logger.info('Found a TS with a single imaginary frequency')
return None
if len(self.imaginary_frequencies) == 0:
logger.error('Transition state optimisation did not return any '
'imaginary frequencies')
return None
# There is more than one imaginary frequency. Will assume that the most
# negative is the correct mode..
for disp_magnitude in [1, -1]:
dis_name_ext = name_ext + '_dis' if disp_magnitude == 1 else name_ext + '_dis2'
atoms, energy, calc = deepcopy(self.atoms), deepcopy(self.energy), deepcopy(self.optts_calc)
self.atoms = get_displaced_atoms_along_mode(self.optts_calc,
mode_number=7,
disp_magnitude=disp_magnitude)
self._run_opt_ts_calc(method=get_hmethod(), name_ext=dis_name_ext)
if len(self.imaginary_frequencies) == 1:
logger.info('Displacement along second imaginary mode '
'successful. Now have 1 imaginary mode')
break
self.optts_calc = calc
self.set_atoms(atoms)
self.energy = energy
self.imaginary_frequencies = self.optts_calc.get_imaginary_freqs()
return None
def find_lowest_energy_ts_conformer(self):
"""Find the lowest energy transition state conformer by performing
constrained optimisations"""
atoms, energy = deepcopy(self.atoms), deepcopy(self.energy)
calc = deepcopy(self.optts_calc)
hmethod = get_hmethod() if Config.hmethod_conformers else None
self.find_lowest_energy_conformer(hmethod=hmethod)
if len(self.conformers) == 1:
logger.warning('Only found a single conformer. '
'Not rerunning TS optimisation')
self.set_atoms(atoms=atoms)
self.energy = energy
self.optts_calc = calc
return None
self.optimise(name_ext='optts_conf')
if self.is_true_ts() and self.energy < energy:
logger.info('Conformer search successful')
else:
logger.warning(f'Transition state conformer search failed '
f'(∆E = {energy - self.energy:.4f} Ha). Reverting')
self.set_atoms(atoms=atoms)
self.energy = energy
self.optts_calc = calc
self.imaginary_frequencies = calc.get_imaginary_freqs()
return None
def is_true_ts(self):
"""Is this TS a 'true' TS i.e. has at least on imaginary mode in the
hessian and is the correct mode"""
if self.energy is None:
logger.warning('Cannot be true TS with no energy')
return False
if len(self.imaginary_frequencies) > 0:
if self.has_correct_imag_mode(calc=self.optts_calc):
logger.info('Found a transition state with the correct '
'imaginary mode & links reactants and products')
return True
return False
def save_ts_template(self, folder_path=Config.ts_template_folder_path):
"""Save a transition state template containing the active bond lengths,
solvent and charge in folder_path
Keyword Arguments:
folder_path (str): folder to save the TS template to
(default: {None})
"""
logger.info(f'Saving TS template for {self.name}')
truncated_graph = get_truncated_active_mol_graph(self.graph,
active_bonds=self.bond_rearrangement.all)
for bond in self.bond_rearrangement.all:
truncated_graph.edges[bond]['distance'] = self.get_distance(*bond)
ts_template = TStemplate(truncated_graph, solvent=self.solvent,
charge=self.charge, mult=self.mult)
ts_template.save_object(folder_path=folder_path)
logger.info('Saved TS template')
return None
def __init__(self, ts_guess):
"""
Transition State
Arguments:
ts_guess (autode.transition_states.ts_guess.TSguess)
"""
super().__init__(atoms=ts_guess.atoms, reactant=ts_guess.reactant,
product=ts_guess.product, name=f'TS_{ts_guess.name}')
self.bond_rearrangement = ts_guess.bond_rearrangement
self.conformers = None
self.optts_calc = None
self.imaginary_frequencies = []
self._update_graph()
# Calculate the energy array
for r in rs:
o_atom, h_atom = water.atoms[:2]
curr_r = water.distance(0, 1)
vector = (h_atom.coord - o_atom.coord) * (r / curr_r - 1)
h_atom.translate(vector)
# Set up and run the calculation
calc = Calculation(name=f'H2O_scan_{r:.2f}',
molecule=water,
method=orca,
keywords=keywords)
calc.run()
# Get the potential energy from the calculation
energy = calc.get_energy()
energies.append(energy)
# Plot the relative energy against the distance. 627.5 kcal mol-1 Ha-1
rel_energies = 627.5 * np.array([e - min(energies) for e in energies])
plt.plot(rs, rel_energies, marker='o', label=dft_name)
# Add labels to the plot and save the figure
plt.ylabel('ΔE / kcal mol$^{-1}$')
plt.xlabel('r / Å')
plt.legend()
plt.savefig('OH_PES_unrelaxed2.png')
o_atom, h_atom = water.atoms[:2]
curr_r = water.get_distance(0, 1) # current O-H distance
# Shift the hydrogen atom to the required distance
# vector = (h_atom.coord - o_atom.coord) / curr_r * (r - curr_r)
vector = (h_atom.coord - o_atom.coord) * (r / curr_r - 1)
h_atom.translate(vector)
# Set up and run the single point energy evaluation
sp = Calculation(name=f'H2O_scan_unrelaxed_{r:.2f}',
molecule=water,
method=xtb,
keywords=xtb.keywords.sp)
sp.run()
sp_energies.append(sp.get_energy())
# Set up the constrained optimisation calculation where the distance
# constraints are given as a dictionary keyed with a tuple of atom indexes
# with the distance as the value
opt = Calculation(name=f'H2O_scan_relaxed_{r:.2f}',
molecule=water,
method=xtb,
keywords=xtb.keywords.low_opt,
distance_constraints={(0, 1): r})
opt.run()
opt_energies.append(opt.get_energy())
# Plot the relative energy against the distance. 627.5 kcal mol-1 Ha-1
rel_sp_energies = 627.5 * np.array([e - min(sp_energies) for e in sp_energies])
plt.plot(rs, rel_sp_energies, marker='o', label='unrelaxed', c='r')
def optimise(molecule,
method,
keywords,
n_cores=None,
cartesian_constraints=None):
"""
Optimise a molecule
:param molecule: (object)
:param method: (autode.ElectronicStructureMethod)
:param keywords: (list(str)) Keywords to use for the electronic structure c
alculation e.g. ['Opt', 'PBE', 'def2-SVP']
:param n_cores: (int) Number of cores to use
:param cartesian_constraints: (list(int)) List of atom ids to constrain
:return:
"""
logger.info('Running an optimisation calculation')
n_cores = Config.n_cores if n_cores is None else int(n_cores)
try:
from autode.calculation import Calculation
from autode.wrappers.XTB import xtb
from autode.wrappers.ORCA import orca
from autode.wrappers.keywords import OptKeywords
except ModuleNotFoundError:
logger.error('autode not found. Calculations not available')
raise RequiresAutodE
if keywords is None:
if method == orca:
keywords = OptKeywords(['LooseOpt', 'PBE', 'D3BJ', 'def2-SVP'])
logger.warning(f'No keywords were set for the optimisation but an'
f' ORCA calculation was requested. '
f'Using {str(keywords)}')
elif method == xtb:
keywords = xtb.keywords.opt
else:
logger.critical('No keywords were set for the optimisation '
'calculation')
raise Exception
else:
# If the keywords are specified as a list convert them to a set of
# OptKeywords, required for autodE
if type(keywords) is list:
keywords = OptKeywords(keywords)
opt = Calculation(name=molecule.name + '_opt',
molecule=molecule,
method=method,
keywords=keywords,
n_cores=n_cores,
cartesian_constraints=cartesian_constraints)
opt.run()
molecule.energy = opt.get_energy()
molecule.set_atoms(atoms=opt.get_final_atoms())
return None
