Overview¶

The MPInterfaces package contains python modules that enable the creation of FireWorks workflows for ligand-slab interface and slab-slab interface calculations with VASP.

The package is structured as python modules (listed in modules section of docs and contained in mpinterfaces folder), examples notebooks (ipython notebooks) and workflows scripts to be used with a locally set up FireWorks database.

The examples folder contains scripts for interface creation and utility post processing and ipython notebook examples and workflow submission scripts.

Apart from these, there are utility scripts for the plotting of band structures, density of states and equations of state.

The example folder “notebooks” also provides example calculations of Pourbaix diagrams, and Phase diagrams based on the method implemented in pymatgen/examples

Capabilities¶

MPInterfaces extends object oriented definitions of materialsproject pymatgen to interfaces and features:

Interface Creation¶

extended object oriented definitions to the pymatgen Slab called “Interface” and an adsorbate extension of pymatgen Molecule called “Ligand” and scripts to create Interfaces.

Example scripts:

create_interface.py : creates an example ligand + slab interface illustrating
ligand_interface.py : creates a list of ligand + slab interfaces with pre-configuration
hetero_interface.py : creates a slab – slab heterostructure interface
nanoparticle.py : creates a nanoparticle based on Wulff Construction

Offline vasp project management¶

The script, mpint, in the scripts folder serves as a management tool for vasp projects, starting from encut, kpoint or other parameter optimization till the solvation calculations. Just define all types of calculations with their corresponding specifications needed for the project in a yaml file and run or rerun calculaitons as required.

it takes 3 arguments: input yaml file, type of calculation and the run mode

example:

` mpint -i naf.yaml -t bulk_calibrate run `

this will read in the specifications for ‘bulk_calibrate’ job from the input yaml file, naf.yaml(in examples folder), and runs the job i.e submits to the PBS que.

Job Monitoring:

Everytime jobs are submitted or its status queried, information such as job ids, job folders etc are written to the calibrate.json(the default name) checkpoint file. This makes it easier to identify job ids and their corresponding job folders. Also provides all job info including converged final energies in one place, in the checkpoint file.

update the calibrate.json(default name) checkpoint file with the final energies(if converged):

` mpint update `

update custom name(example:- step1.json) checkpoint file with the final energies(if converged):

` mpint update step1.json `

update checkpoint files from calibrate runs:

` mpint update step1.json step2.json `

update calibrate.json and rerun jobs(will reset the poscar from the contcar file if one exists) with the given ids:

` mpint update 14692739.moab.ufhpc 14692740.moab.ufhpc `

update calibrate.json and rerun jobs in the given job directory paths:

` mpint update all_poscars/POS/CBr2_294_C1Br2 all_poscars/POS/CoBr2_294_Co1Br2 `

update calibrate.json and rerun jobs in the given folders(job ids are also accepted) with incar or kpoints or que parameters overridden by the parmaters in the override_input.yaml file:

` mpint -i override_input.yaml update all_poscars/POS/CBr2_294_C1Br2 all_poscars/POS/CoBr2_294_Co1Br2 `

High throughput VASP workflows with FireWorks¶

High throughput VASP workflows can be handled by creating your own FireWorks database server. The FireWorks package from the materialsproject is installed at the time of installation. We have defined FireTasks customized to Interface studies as follows:

Calibrate, Measurement and DatabaseSubmit.

Calibrate Task: Defines all tasks that involve the calibration of VASP input parameters. For example, energy cutoff convergence, kpoints convergence,vacuum convergence, thickness convergence in case of slabs Calibrate module defined in calibrate.py provides customized Bulk, Molecule, Slab and Interface objects for the Calibrate Task
Measurement Task: Tasks for VASP jobs dependent on a previous calculation. Currently implemented tasks include Static calculation following a Calibrate Task and Solvation calculation following a Calibrate Task Measurement tasks are linked to calibrate tasks and are capable of copying the necessary files from the previous calculation to restart the calculation. for example, CONTCAR from a Relaxation for a Static calculation, CONTCAR and WAVECAR for a Solvation calculation
Database Task: Submits data to a MongoDB database similar to VasptoDBtaskDrone of pymatgen extended to use Interface specific keys of miller index of slab and ligand chemical formula in .json format

Refer examples/Workflows for example workflow submission scripts.

Example of FireWorks Database Queries¶

After installation and setup of your database according to FireWorks documentation, create your own my_launchpad.yaml using

lpad init

Set the path to this file to LAUNCHPAD_LOC in the fw_config.py found in your installation directory.

TIP: create the my_launchpad.yaml in the directory: ~/.fireworks

Make sure to configure your my_launchpad.yaml file to your local database settings (host name, database name, etc.)

Workflow submission and launching¶

Refer examples folder for simple workflow scripts

-> general submission of workflow to the database:

python workflow_script.py

Fireworks package has some nice utility scripts for launching fireworks and checking job status. If the fireworks package is installed then those scripts are already in your PATH. Some examples are given below:

-> reset database to erase all workflows

CAUTION: Be careful when using the following command as it will erase all workflows from the database:

lpad reset

-> launch a single firework:

rlaunch singleshot -f [fw_id]

-> get all fireworks info:

lpad get_fws

-> get all workflows info:

lpad get_wflows

-> reset database to erase all workflows

CAUTION: Be careful when using the following command as it will erase all workflows from the database:

lpad reset

Workflow query examples¶

-> query by name:

example: lpad get_wflows -n “Ag_100”

-> query by state

lpad get_wflows -s STATE

or

lpad get_wflows -q ‘{“state”:STATE}’ –sort created_on

where STATE = “READY”, “WAITING”, “RUNNING”, “FIZZLED” or “DEFUSED”

-> delete workflows:

example: lpad delete_wflows -n Ag_100

Query for individual Fireworks¶

-> query by state

lpad get_fws -s STATE

or

lpad get_fws -q ‘{“state”:STATE}’ –sort created_on

where STATE = “READY”, “WAITING”, “RUNNING”, “FIZZLED”, “DEFUSED”

-> query fireworks by name or id:

example: lpad get_fws -n “solvation”

lpad get_fws -i 102 -d all

-> re-run firework with id, fw_id. Same as marking the firework as ready

lpad rerun_fws -i fw_id

-> re-run a finished or fizzled firework with updated specs:

example: update the sol_params of the first task of the firework with id 102 lpad rerun_fws -i 102

lpad update_fws -i 102 -u ‘{“_tasks.0.other_params.sol_params.NELECT”:[-1,-0.5,0,0.5,1]}’

Connecting to our local Database¶

The mongo database for job submission(‘fireworks’) is set up on the machine ‘hydrogen’. Please use your own account to connect to the database Contact me (km468@cornell.edu) to create a database account