def minimize_seeds(nprocs=4):
seeds = []
seed_names = []
route_low = "! OPT B97-D3 def2-TZVP ECP{def2-TZVP} Grid7"
extra_section_low = ""
route_high = "! OPT PW6B95 def2-TZVP ECP{def2-TZVP} Grid7"
extra_section_high = ""
for seed in os.listdir("seed"):
seeds.append(files.read_cml("seed/%s" % seed, allow_errors=True, test_charges=False)[0])
seed_names.append(seed.split(".cml")[0])
jobs = []
for i,seed in enumerate(seeds):
charge = sum([a.type.charge for a in seed])
jobs.append( orca.job("seed_%d_low" % i, route_low, atoms=seed, extra_section=extra_section_low, charge=charge, grad=False, queue=QUEUE_TO_RUN_ON, procs=nprocs, sandbox=False) )
for j in jobs: j.wait()
for i,seed in enumerate(seeds):
charge = sum([a.type.charge for a in seed])
jobs.append( orca.job("seed_%d_high" % i, route_high, atoms=[], extra_section=extra_section_high, charge=charge, grad=False, queue=QUEUE_TO_RUN_ON, procs=nprocs, previous="seed_%d_low" % i, sandbox=False) )
for j in jobs: j.wait()
for i,seed in enumerate(seeds):
new_pos = orca.read("seed_%d_high" % i)
if not new_pos.converged:
print("Failed to optimize %s" % seed_names[i])
continue
new_pos = new_pos.atoms
cml_file = files.read_cml("seed/%s" % seed_names[i], allow_errors=True, test_charges=False, return_molecules=True)
j=0
for mol in cml_file:
for k,a in enumerate(mol.atoms):
b = new_pos[j]
a.x, a.y, a.z = b.x, b.y, b.z
j += 1
files.write_cml(cml_file, name="seed/%s_opt" % seed_names[i])
def run_low_level(training_sets_folder="training_set", procs=1, queue=None,
extra_parameters={}):
if not os.path.exists(training_sets_folder):
raise Exception("No training set folder to run.")
frange = [int(a.split('.cml')[0]) for a in os.listdir(training_sets_folder)
if a.endswith(".cml")]
frange.sort()
if len(frange) == 0:
raise Exception("No viable files in training sets folder.")
route = "! B97-D3 def2-TZVP GCP(DFT/TZ) ECP{def2-TZVP} Grid7"
extra_section = ""
running_jobs = []
for i in frange:
atoms = files.read_cml("%s/%d.cml" % (training_sets_folder, i),
allow_errors=True,
test_charges=False,
return_molecules=False,
extra_parameters=extra_parameters)[0]
charge = sum([a.type.charge for a in atoms])
running_jobs.append(orca.job("ts_%d" % i, route,
atoms=atoms,
extra_section=extra_section,
charge=charge,
grad=True,
queue=queue,
procs=procs,
sandbox=False))
return running_jobs
def orca_start_job(spline_NEB, i, state, charge, procs, queue, initial_guess,
extra_section, mem, priority):
"""
A method for submitting a single point calculation using Orca for spline_NEB
calculations.
**Parameters**
spline_NEB: :class:`spline_NEB`
A spline_NEB container holding the main spline_NEB simulation
i: *int*
The index corresponding to which image on the frame is to be
simulated.
state: *list,* :class:`structures.Atom`
A list of atoms describing the image on the frame associated with
index *i*.
charge: *int*
Charge of the system.
procs: *int*
The number of processors to use during calculations.
queue: *str*
Which queue to submit the simulation to (this is queueing system
dependent).
initial_guess: *str*
The name of a previous simulation for which we can read in a
hessian.
extra_section: *str*
Extra settings for this DFT method.
mem: *int*
How many MegaBytes (MB) of memory you have available per core.
**Returns**
orca_job: :class:`jobs.Job`
A job container holding the orca simulation.
"""
if spline_NEB.step > 0:
previous = '%s-%d-%d' % (spline_NEB.name, spline_NEB.step - 1, i)
else:
if initial_guess:
if hasattr(initial_guess, '__iter__'):
previous = initial_guess[i]
else:
previous = initial_guess
else:
previous = None
return orca.job('%s-%d-%d' % (spline_NEB.name, spline_NEB.step, i),
spline_NEB.theory,
state,
charge=charge,
extra_section=extra_section,
grad=True,
procs=procs,
queue=queue,
previous=previous,
mem=mem,
priority=spline_NEB.priority)
def run_high_level(training_sets_folder="training_set",
procs=1,
queue=None,
extra_parameters={}):
if not os.path.exists(training_sets_folder):
raise Exception("No training set folder to run.")
frange = [
int(a.split('.cml')[0]) for a in os.listdir(training_sets_folder)
if a.endswith(".cml")
]
frange.sort()
if len(frange) == 0:
raise Exception("No viable files in training sets folder.")
route = "! PW6B95 def2-TZVP GCP(DFT/TZ) ECP{def2-TZVP} Grid7"
extra_section = ""
running_jobs = []
previous_failed = []
for i in frange:
atoms = files.read_cml("%s/%d.cml" % (training_sets_folder, i),
allow_errors=True,
test_charges=False,
return_molecules=False,
extra_parameters=extra_parameters)[0]
charge = sum([a.type.charge for a in atoms])
prev_converged = orca.read("ts_%d" % i).converged
if prev_converged:
running_jobs.append(
orca.job("ts_%d_high" % i,
route,
atoms=[],
extra_section=extra_section,
charge=charge,
grad=True,
queue=queue,
procs=procs,
previous="ts_%d" % i,
sandbox=False))
else:
previous_failed.append(i)
return running_jobs, previous_failed
def run_high_level():
if not os.path.exists("training_sets"):
raise Exception("No training set folder to run.")
frange = [int(a.split('.cml')[0]) for a in os.listdir("training_sets") if a.endswith(".cml")]
frange.sort()
if len(frange) == 0:
raise Exception("No viable files in training sets folder.")
route = "! PW6B95 def2-TZVP GCP(DFT/TZ) ECP{def2-TZVP} Grid7"
extra_section = ""
running_jobs = []
previous_failed = []
for i in frange:
atoms = files.read_cml("training_sets/%d.cml" % i, allow_errors=True, test_charges=False, return_molecules=False)[0]
charge = sum([a.type.charge for a in atoms])
prev_converged = orca.read("ts_%d" % i).converged
if prev_converged:
running_jobs.append( orca.job("ts_%d_high" % i, route, atoms=[], extra_section=extra_section, charge=charge, grad=True, queue=QUEUE_TO_RUN_ON, procs=QUEUE_PROCS, previous="ts_%d" % i, sandbox=False) )
else:
previous_failed.append(i)
return running_jobs, previous_failed
def minimize_seeds(procs=4, queue=None, extra_parameters={}):
"""
A function to optimize the geometry of the supplied seeds in the seed
directory. Each optimized structure is then added to the seed directory
under the name "previous_name_opt.cml".
**Parameters**
procs: *int, optional*
How many processors to use for this.
queue: *str, optional*
What queue to run the simulation on.
"""
seeds = []
seed_names = []
route_low = "! OPT B97-D3 def2-TZVP ECP{def2-TZVP} Grid7"
extra_section_low = ""
route_high = "! OPT PW6B95 def2-TZVP ECP{def2-TZVP} Grid7"
extra_section_high = ""
for seed in os.listdir("seed"):
seeds.append(
files.read_cml("seed/%s" % seed,
allow_errors=True,
test_charges=False,
extra_parameters=extra_parameters)[0])
seed_names.append(seed.split(".cml")[0])
jobs = []
for i, seed in enumerate(seeds):
charge = sum([a.type.charge for a in seed])
jobs.append(
orca.job("seed_%d_low" % i,
route_low,
atoms=seed,
extra_section=extra_section_low,
charge=charge,
grad=False,
queue=queue,
procs=procs,
sandbox=False))
for j in jobs:
j.wait()
for i, seed in enumerate(seeds):
charge = sum([a.type.charge for a in seed])
jobs.append(
orca.job("seed_%d_high" % i,
route_high,
atoms=[],
extra_section=extra_section_high,
charge=charge,
grad=False,
queue=queue,
procs=procs,
previous="seed_%d_low" % i,
sandbox=False))
for j in jobs:
j.wait()
for i, seed in enumerate(seeds):
new_pos = orca.read("seed_%d_high" % i)
if not new_pos.converged:
print("Failed to optimize %s" % seed_names[i])
continue
new_pos = new_pos.atoms
cml_file = files.read_cml("seed/%s" % seed_names[i],
allow_errors=True,
test_charges=False,
return_molecules=True,
extra_parameters=extra_parameters)
j = 0
for mol in cml_file:
for k, a in enumerate(mol.atoms):
b = new_pos[j]
a.x, a.y, a.z = b.x, b.y, b.z
j += 1
files.write_cml(cml_file, name="seed/%s_opt" % seed_names[i])
def orca_start_job(NEB,
i,
state,
charge,
multiplicity,
procs,
queue,
initial_guess,
extra_section,
mem,
priority,
extra_keywords={}):
"""
A method for submitting a single point calculation using Orca for NEB
calculations.
**Parameters**
NEB: :class:`NEB`
An NEB container holding the main NEB simulation
i: *int*
The index corresponding to which image on the frame is to be
simulated.
state: *list,* :class:`structures.Atom`
A list of atoms describing the image on the frame associated with
index *i*.
charge: *int*
Charge of the system.
multiplicity: *int*
Multiplicity of the system.
procs: *int*
The number of processors to use during calculations.
queue: *str*
Which queue to submit the simulation to (this is queueing system
dependent).
initial_guess: *str*
The name of a previous simulation for which we can read in a
hessian.
extra_section: *str*
Extra settings for this DFT method.
mem: *int*
How many MegaBytes (MB) of memory you have available per core.
priority: *int*
Whether to submit to NBS with a given priority
extra_keywords: *dict, optional*
Specify extra keywords beyond the defaults.
**Returns**
orca_job: :class:`jobs.Job`
A job container holding the orca simulation.
"""
orca4 = True
if "orca4" in extra_keywords:
orca4 = extra_keywords["orca4"]
NEB.calls_to_force += 1
if NEB.step > 0:
previous = '%s-%d-%d' % (NEB.name, NEB.step - 1, i)
else:
if initial_guess:
if hasattr(initial_guess, '__iter__'):
previous = initial_guess[i]
else:
previous = initial_guess
else:
previous = None
return orca.job('%s-%d-%d' % (NEB.name, NEB.step, i),
NEB.theory,
state,
charge=charge,
multiplicity=multiplicity,
extra_section=extra_section,
grad=True,
procs=procs,
queue=queue,
previous=previous,
mem=mem,
priority=NEB.priority,
orca4=orca4)
def generate_lead_halide_cation(halide, cation, ion="Pb", run_opt=True):
cml_path = fpl_constants.cml_dir
# Check if system exists
fname = reduce_to_name(ion, halide, cation)
if not cml_path.endswith("/"):
cml_path += "/"
if os.path.exists(cml_path + fname + ".cml"):
print("Found system in cml folder, returning system")
system = structures.Molecule(
files.read_cml(cml_path + fname + ".cml",
test_charges=False,
allow_errors=True)[0])
return system
def vdw(y):
return PERIODIC_TABLE[units.elem_s2i(y)]['vdw_r']
# Get the PbX3 system
PbX3 = generate_lead_halide(halide, ion=ion)
# Get the cation from the cml file
atoms, bonds, _, _ = files.read_cml(cml_path + cation + ".cml",
test_charges=False,
allow_errors=True)
system = structures.Molecule(atoms)
# Align along X axis
system.atoms = geometry.align_centroid(system.atoms)[0]
# Rotate to Z axis
# NOTE! In case of FA, we want flat so only translate to origin instead
# NOTE! We have exactly 3 cations we observe: Cs, MA, FA. If 2 N, then FA
elems = [a.element for a in system.atoms]
if elems.count("N") == 2:
system.translate(system.get_center_of_mass())
else:
R = geometry.rotation_matrix([0, 1, 0], 90, units="deg")
system.rotate(R)
# If N and C in system, ensure N is below C (closer to Pb)
if "N" in elems and "C" in elems:
N_index = [i for i, a in enumerate(system.atoms)
if a.element == "N"][0]
C_index = [i for i, a in enumerate(system.atoms)
if a.element == "C"][0]
if system.atoms[N_index].z > system.atoms[C_index].z:
# Flip if needed
R = geometry.rotation_matrix([0, 1, 0], 180, units="deg")
system.rotate(R)
# Offset system so lowest point is at 0 in the z dir
z_offset = min([a.z for a in system.atoms]) * -1
system.translate([0, 0, z_offset])
# Add to the PbX3 system with an offset of vdw(Pb)
system.translate([0, 0, vdw(ion)])
system.atoms += PbX3.atoms
# Run a geometry optimization of this system
if run_opt:
PbXY = orca.job(fname,
fpl_constants.default_routes[0],
atoms=system.atoms,
extra_section=fpl_constants.extra_section,
queue="batch",
procs=2)
PbXY.wait()
new_pos = orca.read(fname).atoms
for a, b in zip(system.atoms, new_pos):
a.x, a.y, a.z = [b.x, b.y, b.z]
# Set OPLS types
for a in system.atoms:
if a.element in [ion, "Cl", "Br", "I"]:
a.type = fpl_constants.atom_types[a.element]
a.type_index = a.type["index"]
# Write cml file so we don't re-generate, and return system
files.write_cml(system, bonds=bonds, name=cml_path + fname + ".cml")
return system
def minimize_seeds(procs=4, queue=None, extra_parameters={}):
"""
A function to optimize the geometry of the supplied seeds in the seed
directory. Each optimized structure is then added to the seed directory
under the name "previous_name_opt.cml".
**Parameters**
procs: *int, optional*
How many processors to use for this.
queue: *str, optional*
What queue to run the simulation on.
"""
seeds = []
seed_names = []
route_low = "! OPT B97-D3 def2-TZVP ECP{def2-TZVP} Grid7"
extra_section_low = ""
route_high = "! OPT PW6B95 def2-TZVP ECP{def2-TZVP} Grid7"
extra_section_high = ""
for seed in os.listdir("seed"):
seeds.append(files.read_cml("seed/%s" % seed,
allow_errors=True,
test_charges=False,
extra_parameters=extra_parameters)[0])
seed_names.append(seed.split(".cml")[0])
jobs = []
for i, seed in enumerate(seeds):
charge = sum([a.type.charge for a in seed])
jobs.append(orca.job("seed_%d_low" % i, route_low,
atoms=seed,
extra_section=extra_section_low,
charge=charge,
grad=False,
queue=queue,
procs=procs,
sandbox=False))
for j in jobs:
j.wait()
for i, seed in enumerate(seeds):
charge = sum([a.type.charge for a in seed])
jobs.append(orca.job("seed_%d_high" % i, route_high,
atoms=[],
extra_section=extra_section_high,
charge=charge,
grad=False,
queue=queue,
procs=procs,
previous="seed_%d_low" % i,
sandbox=False))
for j in jobs:
j.wait()
for i, seed in enumerate(seeds):
new_pos = orca.read("seed_%d_high" % i)
if not new_pos.converged:
print("Failed to optimize %s" % seed_names[i])
continue
new_pos = new_pos.atoms
cml_file = files.read_cml("seed/%s" % seed_names[i],
allow_errors=True,
test_charges=False,
return_molecules=True,
extra_parameters=extra_parameters)
j = 0
for mol in cml_file:
for k, a in enumerate(mol.atoms):
b = new_pos[j]
a.x, a.y, a.z = b.x, b.y, b.z
j += 1
files.write_cml(cml_file, name="seed/%s_opt" % seed_names[i])