def run_create(self):
"""
Launch the first, "create" sub-process.
"""
self.report(f"Running create process '{self.inputs.create_process.value}'")
create_process_class = load_object(self.inputs.create_process.value)
return ToContext(
create_process=self.run_or_submit(create_process_class, **self.inputs.create)
)
def fp_run(self):
"""
Run the first-principles calculation workflows.
"""
self.report('Submitting reference_bands workflow.')
reference_bands = self.submit(
load_object(self.inputs.reference_bands_workflow),
**ChainMap(
self.inputs['reference_bands'],
self.exposed_inputs(ReferenceBandsBase,
namespace='reference_bands')))
self.report('Submitting wannier_input workflow.')
wannier_input = self.submit(
load_object(self.inputs.wannier_input_workflow),
**ChainMap(
self.inputs['wannier_input'],
self.exposed_inputs(WannierInputBase,
namespace='wannier_input')))
return ToContext(reference_bands=reference_bands,
wannier_input=wannier_input)
def fp_run(self):
"""
Runs the first-principles calculation workflow.
"""
self.report("Starting DFT workflows.")
return ToContext(fp_run=self.submit(
load_object(self.inputs.fp_run_workflow),
**ChainMap(
self.inputs.fp_run,
self.exposed_inputs(FirstPrinciplesRunBase,
namespace='fp_run'),
)))
def evaluate_bands(self):
"""
Add the tight-binding model to the outputs and run the evaluation workflow.
"""
self.report("Adding tight-binding model to output.")
tb_model = self.ctx.tbextraction_calc.outputs.tb_model
self.out('tb_model', tb_model)
self.report("Running model evaluation.")
return ToContext(model_evaluation_wf=self.submit(
load_object(self.inputs.model_evaluation_workflow),
tb_model=tb_model,
**ChainMap(
self.inputs.model_evaluation,
self.exposed_inputs(ModelEvaluationBase),
)))
def launch_evaluations(self):
"""
Create evaluations for the current iteration step.
"""
self.report('Launching pending evaluations.')
with self.optimizer() as opt:
evals = {}
evaluate_process = load_object(self.inputs.evaluate_process.value)
for idx, inputs in opt.create_inputs().items():
self.report('Launching evaluation {}'.format(idx))
inputs_merged = ChainMap(inputs,
self.inputs.get('evaluate', {}))
if is_process_function(evaluate_process):
_, node = run_get_node(evaluate_process, **inputs_merged)
else:
node = self.submit(evaluate_process, **inputs_merged)
evals[self.eval_key(idx)] = node
self.indices_to_retrieve.append(idx)
return self.to_context(**evals)
def run_evaluate(self):
"""
Retrieve outputs of the "create" sub-process, and launch the
"evaluate" sub-process.
"""
create_process_outputs = get_outputs_dict(self.ctx.create_process)
self.out('create', create_process_outputs)
if not self.ctx.create_process.is_finished_ok:
return self.exit_codes.ERROR_CREATE_PROCESS_FAILED
self.report(f"Running evaluate process '{self.inputs.evaluate_process.value}'")
evaluate_process_class = load_object(self.inputs.evaluate_process.value)
output_input_mapping = self.inputs.output_input_mapping.get_dict()
created_inputs = {
in_key: create_process_outputs[out_key]
for out_key, in_key in output_input_mapping.items() if out_key in create_process_outputs
}
return ToContext(
evaluate_process=self.
run_or_submit(evaluate_process_class, **self.inputs.evaluate, **created_inputs)
)
def run_evaluate(self):
"""
Runs the model evaluation workflow.
"""
tb_model = self.ctx.tbextraction_calc.outputs.tb_model
self.report("Adding tight-binding model to output.")
self.out('tb_model', tb_model)
# slice reference bands if necessary
reference_bands = self.ctx.fp_run.outputs.bands
slice_reference_bands = self.inputs.get('slice_reference_bands', None)
if slice_reference_bands is not None:
reference_bands = slice_bands_inline(
bands=reference_bands, slice_idx=slice_reference_bands)
self.report('Starting model evaluation workflow.')
return ToContext(model_evaluation_wf=self.submit(
load_object(self.inputs.model_evaluation_workflow),
tb_model=tb_model,
reference_bands=reference_bands,
**ChainMap(
self.inputs.model_evaluation,
self.exposed_inputs(ModelEvaluationBase),
)))
def run_process(self):
"""
Merge the inputs namespace and added inputs, and launch the
sub-process.
"""
self.report("Merging inputs for the sub-process.")
if isinstance(self.inputs.added_input_keys, orm.Str):
added_input_keys = [self.inputs.added_input_keys.value]
if not isinstance(self.inputs.added_input_values, orm.BaseType):
raise InputValidationError(
"When 'added_input_keys' is given as 'Str', 'added_input_values'"
" must be a 'BaseType' instance.")
added_input_values = [self.inputs.added_input_values.value]
else:
added_input_keys = self.inputs.added_input_keys.get_list()
if not isinstance(self.inputs.added_input_values, orm.List):
raise InputValidationError(
"When 'added_input_keys' is given as 'List', 'added_input_values'"
" must also be a 'List'.")
added_input_values = self.inputs.added_input_values.get_list()
if len(added_input_values) != len(added_input_keys):
raise InputValidationError(
"Lengths of 'added_input_values' and 'added_input_keys' do not match."
)
inputs = AttributeDict(self.inputs.inputs)
def _get_or_create_sub_dict(in_dict, name):
try:
return in_dict[name]
except KeyError:
res = {}
in_dict[name] = res
return res
def _get_or_create_port(in_attr_dict, name):
try:
return getattr(in_attr_dict, name)
except AttributeError:
res = AttributeDict()
setattr(in_attr_dict, name, res)
return res
for key, value in zip(added_input_keys, added_input_values):
full_port_path, *full_attr_path = key.split(':')
*port_path, port_name = full_port_path.split('.')
namespace = reduce(_get_or_create_port, port_path, inputs)
if not full_attr_path:
res_value = to_aiida_type(value)
else:
assert len(full_attr_path) == 1
# Get or create the top-level dictionary.
try:
res_dict = getattr(namespace, port_name).get_dict()
except AttributeError:
res_dict = {}
*sub_dict_path, attr_name = full_attr_path[0].split('.')
sub_dict = reduce(_get_or_create_sub_dict, sub_dict_path,
res_dict)
sub_dict[attr_name] = value
res_value = orm.Dict(dict=res_dict).store()
setattr(namespace, port_name, res_value)
self.report("Launching the sub-process.")
return ToContext(sub_process=self.run_or_submit(
load_object(self.inputs.sub_process.value), **inputs))
def engine(self):
return load_object(self.inputs.engine.value)
