This tutorial will show you how Lobster calculations can be run with atomate (https://atomate.org/) and pymatgen (https://pymatgen.org/). Written by Janine George (janine.george@bam.de) and Aakash Naik (aakash.naik@bam.de).

Please cite our latest Lobster publication if you make use of this tutorial:

R. Nelson, C. Ertural, J. George, V. L. Deringer, G. Hautier, R. Dronskowski, J Comput Chem. 2020, 41, 1931.[PDF]

First load all important classes, methods

from pymatgen.ext.matproj import MPRester
from atomate.vasp.powerups import add_additional_fields_to_taskdocs, set_execution_options
from atomate.vasp.workflows.base.lobster import get_wf_lobster_test_basis
from fireworks import LaunchPad

Get a structure from the Materials Project that you would like to run (or from elsewhere)

mpid="mp-573697"

a = MPRester()
struct = a.get_structure_by_material_id(mpid)

This workflow performs following steps sequentially:

1. Perform an geometry optimization of the retrieved structure
2. Static vasp calculation (Computes wavecar needed for lobster runs)
3. Test all possible basis functions (Lobster runs)
4. Finally will delete the wavecars

(We use PBE_54 pseudopotentials by default!)

For this tutorial we would be using get_wf_lobster_test_basis workflow from atomate.

This workflow at present is flexible and can be tuned as per users needs. However, there are some limitations and further options will be added soon in near future. Following are the list of arguments applicable to get_wf_lobster_test_basis workflow:

• structure (Structure): structure object that will be used during the run
• calculation_type (str): only "standard" is implemented so far
• delete_all_wavecars (bool): all wavecars wil be deleted if True
• c (dict): specifications for wf, e.g. VASP_CMD, LOBSTER_CMD etc.
• address_max_basis (str): address to yaml file including maximum basis set (otherwise predefined file)
• address_min_basis (str): address to yaml file including minimum basis set (otherwise predefined file)
• user_lobsterin_settings (dict): change lobsterin settings here
• user_incar_settings (dict): change incar settings with this dict, only for Lobster calc and not optimization
• user_kpoints_settings (dict): change kpoint settings with this dict, only for Lobster calc and not optimization
• isym (int): isym setting during the VASP calculation, currently lobster can only deal with isym=-1 and isym=0
• additional_outputs (list): list of additional files to be stored in the results DB. They will be stored as files in gridfs. Examples are:"ICOHPLIST.lobster" or "DOSCAR.lobster". Note that the file name should be given with the full name and the correct capitalization.
• additional_optimization (bool): determines if an optmization is performed
• user_incar_settings_optimization (dict): change incar settings with this dict for optimization
• user_kpoints_settings_optimization (dict): change incar settings with this dict for optimization

Next code block shows an example of using this workflow to add calculation to DB

wf = get_wf_lobster_test_basis(structure=struct, 
                               user_kpoints_settings={"grid_density": 6000},
                               user_incar_settings={"LCHARG": False, "ISPIN": 1, "MAGMOM": None},
                               delete_all_wavecars=True,
                               additional_optimization=True,
                               user_incar_settings_optimization={"ISMEAR": 1, "EDIFFG": 1e-5, 
                                                                 "EDIFF": 1e-6, "NELM": 3000, 
                                                                 "ISPIN": 1,"MAGMOM": None},
                               user_kpoints_settings_optimization={"grid_density": 6000})

Add additional information to the database (e.g., name of the material)

update_dict={"material-id": mpid}

#This is a powerup to add the material-id to the database
wf=add_additional_fields_to_taskdocs(wf,update_dict=update_dict, task_name_constraint='VaspToDb')
wf=add_additional_fields_to_taskdocs(wf,update_dict=update_dict, task_name_constraint='LobsterRunToDb')

Snapshot of generated fireworks workflow from its webgui interface from above code snippet

It is better to use different "workers" for different job types, we will use a powerup to set this.

The standard one should be used for VASP. Then, you need one for lobster and for deleting the WAVECARs. The name of those workers should be corresponding to "fworker_name".

Here, it would be "manneback_lobster" and "manneback_delete".

(One can rename this tags to own liking, just need to make sure you change the corresponding fworker_name in my_fworker_lobster.yaml and my_fworker_delete.yaml respectively)

Sample my_fworker_lobster.yaml file looks like this

name: 'manneback_lobster'
category: ''
query: '{}'
env:
   db_file: /path/to/db.json
   vasp_cmd:  mpirun -np 48  /path/to/vasp/
   lobster_cmd: /path/to/Lobster/
   scratch_dir: null
   auto_npar: True

wf=set_execution_options(wf, fworker_name="manneback_lobster", category=None, fw_name_constraint="lobster_calculation", task_name_constraint=None)
wf=set_execution_options(wf, fworker_name="manneback_delete", category=None, fw_name_constraint="DelteWavecar", task_name_constraint=None)

Adds workflows to launchpad (lpad)

lpad = LaunchPad.auto_load() # loads this based on the FireWorks configuration
lpad.add_wf(wf)

How to start the jobs:

You have to write my_fworker_lobster.yaml and my_fworker_delete.yaml files and correspnding queueadapter files that are called "my_qadapter_lobster.yaml" and "my_qadapter_delete.yaml" here.

And then, use the following commands in your shell:

qlaunch -r rapidfire
qlaunch -r --fworker_file ~/.fireworks/my_fworker_lobster.yaml --queueadapter_file ~/.fireworks/my_qadapter_lobster.yaml rapidfire
qlaunch -r --fworker_file ~/.fireworks/my_fworker_delete.yaml --queueadapter_file ~/.fireworks/my_qadapter_delete.yaml rapidfire