About the ISMEAR and SIGMA tags used in your examples.
#2
Comments
|
Hi Hongyi, For ISMEAR, I typically follow the VASP manual's recommendations. When calculating the Density of States (DOS), I use the tetrahedron method with ISMEAR = -5, which allows for DOS calculations without smearing. In most cases, Gaussian smearing with ISMEAR = 0 is effective for a wide range of systems when needed. For SIGMA (the smearing width), it should be chosen in conjunction with the k-point mesh: a denser k-point grid allows for a smaller SIGMA, while a coarser grid may need a higher value. Additionally, this choice may depend on the type of compound. For example, metals often require a denser k-point mesh paired with a larger SIGMA value. Also, if you're performing ionic relaxation, ensure that the forces have converged. Best, |
|
IMHO, a more robust way is to use tools like custodian to tackle such things automatically, as shown below: Test 1:import numpy as np
from jobflow import run_locally
from pymatgen.symmetry.analyzer import SpacegroupAnalyzer
from atomate2.vasp.flows.elastic import ElasticMaker
from atomate2.vasp.powerups import (
update_user_incar_settings,
update_user_potcar_functional,
update_user_kpoints_settings,
)
from pymatgen.io.vasp.inputs import Kpoints
from pymatgen.core import Structure
from mp_api.client import MPRester
from atomate2.vasp.powerups import add_metadata_to_flow
#https://github.com/materialsproject/atomate2/issues/1014#issuecomment-2454955891
from atomate2.vasp.flows.mp import MPGGADoubleRelaxMaker, MPGGAStaticMaker
user_potcar_functional = "PBE_64"
material_id = "mp-126"
with MPRester() as mpr:
# Obtain elastic data and the related task IDs.
elasticity_doc = mpr.materials.elasticity.search(material_ids=[material_id])
opt_id = elasticity_doc[0].fitting_data.optimization_task.string
first_deform_id = str(elasticity_doc[0].fitting_data.deformation_tasks[0])
opt_doc = mpr.materials.tasks.search([opt_id],
fields=["input", "orig_inputs", "calcs_reversed"])
deform_dep_doc = mpr.materials.tasks.search([first_deform_id],
fields=["input", "orig_inputs", "calcs_reversed"])
structure = mpr.materials.get_structure_by_material_id(material_id)
structure = SpacegroupAnalyzer(structure).get_conventional_standard_structure()
opt_kpoints_settings = opt_doc[0].calcs_reversed[0].input.kpoints
deform_kpoints_settings = deform_dep_doc[0].calcs_reversed[0].input.kpoints
opt_incar_settings = opt_doc[0].input.incar
deform_incar_settings = deform_dep_doc[0].input.incar
# The purpose of this test is to solely assess the impact of using the MP's following incar settings on the final result.
keys_to_extract = {'ISMEAR', 'SIGMA'}
opt_incar_settings1 = {k: v for k, v in opt_incar_settings.items() if k in keys_to_extract}
deform_incar_settings1 = {k: v for k, v in deform_incar_settings.items() if k in keys_to_extract}
print(opt_incar_settings1)
print(deform_incar_settings1)
flow = ElasticMaker(
bulk_relax_maker=MPGGADoubleRelaxMaker(), # 使用MP的双重优化设置
elastic_relax_maker=MPGGAStaticMaker(), # 使用MP的静态计算设置
).make(structure, conventional=True)
flow = update_user_potcar_functional(flow, user_potcar_functional)
# flow = update_user_kpoints_settings(flow, opt_kpoints_settings, name_filter="MP GGA relax")
# flow = update_user_kpoints_settings(flow, deform_kpoints_settings, name_filter="MP GGA static")
#flow = update_user_incar_settings(flow, opt_incar_settings1, name_filter="MP GGA relax" )
#flow = update_user_incar_settings(flow, deform_incar_settings1, name_filter="MP GGA static" )
# Run the flow
responses = run_locally(flow, create_folders=True, ensure_success=True)
elastic_output = responses[flow.jobs[-1].uuid][1].output
matrix = np.array(elastic_output.elastic_tensor.ieee_format)
np.set_printoptions(precision=4, suppress=True)
print(matrix)The result: Test 2:In this test, we use the same kpoints settings as those used by MP: import numpy as np
from jobflow import run_locally
from pymatgen.symmetry.analyzer import SpacegroupAnalyzer
from atomate2.vasp.flows.elastic import ElasticMaker
from atomate2.vasp.powerups import (
update_user_incar_settings,
update_user_potcar_functional,
update_user_kpoints_settings,
)
from pymatgen.io.vasp.inputs import Kpoints
from pymatgen.core import Structure
from mp_api.client import MPRester
from atomate2.vasp.powerups import add_metadata_to_flow
#https://github.com/materialsproject/atomate2/issues/1014#issuecomment-2454955891
from atomate2.vasp.flows.mp import MPGGADoubleRelaxMaker, MPGGAStaticMaker
user_potcar_functional = "PBE_64"
material_id = "mp-126"
with MPRester() as mpr:
# Obtain elastic data and the related task IDs.
elasticity_doc = mpr.materials.elasticity.search(material_ids=[material_id])
opt_id = elasticity_doc[0].fitting_data.optimization_task.string
first_deform_id = str(elasticity_doc[0].fitting_data.deformation_tasks[0])
opt_doc = mpr.materials.tasks.search([opt_id],
fields=["input", "orig_inputs", "calcs_reversed"])
deform_dep_doc = mpr.materials.tasks.search([first_deform_id],
fields=["input", "orig_inputs", "calcs_reversed"])
structure = mpr.materials.get_structure_by_material_id(material_id)
structure = SpacegroupAnalyzer(structure).get_conventional_standard_structure()
opt_kpoints_settings = opt_doc[0].calcs_reversed[0].input.kpoints
deform_kpoints_settings = deform_dep_doc[0].calcs_reversed[0].input.kpoints
opt_incar_settings = opt_doc[0].input.incar
deform_incar_settings = deform_dep_doc[0].input.incar
# The purpose of this test is to solely assess the impact of using the MP's following incar settings on the final result.
keys_to_extract = {'ISMEAR', 'SIGMA'}
opt_incar_settings1 = {k: v for k, v in opt_incar_settings.items() if k in keys_to_extract}
deform_incar_settings1 = {k: v for k, v in deform_incar_settings.items() if k in keys_to_extract}
print(opt_incar_settings1)
print(deform_incar_settings1)
flow = ElasticMaker(
bulk_relax_maker=MPGGADoubleRelaxMaker(), # 使用MP的双重优化设置
elastic_relax_maker=MPGGAStaticMaker(), # 使用MP的静态计算设置
).make(structure, conventional=True)
flow = update_user_potcar_functional(flow, user_potcar_functional)
flow = update_user_kpoints_settings(flow, opt_kpoints_settings, name_filter="MP GGA relax")
flow = update_user_kpoints_settings(flow, deform_kpoints_settings, name_filter="MP GGA static")
#flow = update_user_incar_settings(flow, opt_incar_settings1, name_filter="MP GGA relax" )
#flow = update_user_incar_settings(flow, deform_incar_settings1, name_filter="MP GGA static" )
# Run the flow
responses = run_locally(flow, create_folders=True, ensure_success=True)
elastic_output = responses[flow.jobs[-1].uuid][1].output
matrix = np.array(elastic_output.elastic_tensor.ieee_format)
np.set_printoptions(precision=4, suppress=True)
print(matrix)The result: Here are some more tests I have conducted:Test 3:With the following settings: user_potcar_functional = "PBE_64"
keys_to_extract = {'ALGO'}The corresponding complete python code: import numpy as np
from jobflow import run_locally
from pymatgen.symmetry.analyzer import SpacegroupAnalyzer
from atomate2.vasp.flows.elastic import ElasticMaker
from atomate2.vasp.powerups import (
update_user_incar_settings,
update_user_potcar_functional,
update_user_kpoints_settings,
)
from pymatgen.io.vasp.inputs import Kpoints
from pymatgen.core import Structure
from mp_api.client import MPRester
from atomate2.vasp.powerups import add_metadata_to_flow
#https://github.com/materialsproject/atomate2/issues/1014#issuecomment-2454955891
from atomate2.vasp.flows.mp import MPGGADoubleRelaxMaker, MPGGAStaticMaker
user_potcar_functional = "PBE_64"
material_id = "mp-126"
with MPRester() as mpr:
# Obtain elastic data and the related task IDs.
elasticity_doc = mpr.materials.elasticity.search(material_ids=[material_id])
opt_id = elasticity_doc[0].fitting_data.optimization_task.string
first_deform_id = str(elasticity_doc[0].fitting_data.deformation_tasks[0])
opt_doc = mpr.materials.tasks.search([opt_id],
fields=["input", "orig_inputs", "calcs_reversed"])
deform_dep_doc = mpr.materials.tasks.search([first_deform_id],
fields=["input", "orig_inputs", "calcs_reversed"])
structure = mpr.materials.get_structure_by_material_id(material_id)
structure = SpacegroupAnalyzer(structure).get_conventional_standard_structure()
opt_kpoints_settings = opt_doc[0].calcs_reversed[0].input.kpoints
deform_kpoints_settings = deform_dep_doc[0].calcs_reversed[0].input.kpoints
opt_incar_settings = opt_doc[0].input.incar
deform_incar_settings = deform_dep_doc[0].input.incar
# The purpose of this test is to solely assess the impact of using the MP's following incar settings on the final result.
keys_to_extract = {'ALGO'}
opt_incar_settings1 = {k: v for k, v in opt_incar_settings.items() if k in keys_to_extract}
deform_incar_settings1 = {k: v for k, v in deform_incar_settings.items() if k in keys_to_extract}
print(opt_incar_settings1)
print(deform_incar_settings1)
flow = ElasticMaker(
bulk_relax_maker=MPGGADoubleRelaxMaker(), # 使用MP的双重优化设置
elastic_relax_maker=MPGGAStaticMaker(), # 使用MP的静态计算设置
).make(structure, conventional=True)
flow = update_user_potcar_functional(flow, user_potcar_functional)
flow = update_user_kpoints_settings(flow, opt_kpoints_settings, name_filter="MP GGA relax")
flow = update_user_kpoints_settings(flow, deform_kpoints_settings, name_filter="MP GGA static")
flow = update_user_incar_settings(flow, opt_incar_settings1, name_filter="MP GGA relax" )
flow = update_user_incar_settings(flow, deform_incar_settings1, name_filter="MP GGA static" )
# Run the flow
responses = run_locally(flow, create_folders=True, ensure_success=True)
elastic_output = responses[flow.jobs[-1].uuid][1].output
matrix = np.array(elastic_output.elastic_tensor.ieee_format)
np.set_printoptions(precision=4, suppress=True)
print(matrix) The result: For simplicity, in the following testings, I only list the different settings: Test 4:The settings: user_potcar_functional = "PBE"
keys_to_extract = {'ALGO'}The result: Test 5:The settings: user_potcar_functional = "PBE"
keys_to_extract = {'ALGO', 'ISMEAR', 'SIGMA'}The result: Therefore, we can summarize as follows:
For the related discussion, see #1 (comment). |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Hi there,
I noticed in the examples of this package, the
ISMEARandSIGMAtags are set as follows:As we all know, setting up these two tags for different systems can be tricky. So, I would like to know your strategy for reasonably setting these tags.
Regards,
Zhao
The text was updated successfully, but these errors were encountered: