MPmorph

MPmorph is a collection of tools to run and analyze ab-initio molecular dynamics (AIMD) calculations run with VASP, and is currently under development. It relies heavily on tools developed by the Materials Project (pymatgen, custodian, fireworks) and atomate.

MPmorph provides:

Infrastructure for dynamic VASP MD workflows:
- Tools to create dynamic MD workflows using atomate
- E.g. generation of new MD runs based on the given criterion (currently pressure, for liquid/amorphous phase density estimation)
Tools for statistical analysis of:
- Static observables:
  - Radial distribution functions
  - Coordination numbers
  - Voronoi analysis
  - Polyhedra connectivities
  - Thermodynamic quantities
- Diffusion coefficients:
  - Robust calculation of diffusion coefficients (D) and activation energies (Q).
  - Rigorous error analysis for D and Q.

Installation

Before installing mpmorph, install the latest version of pymatgen, custodian, fireworks) and atomate

clone the repository to your computer and install using

python setup.py develop

If you wish to make amorphous structures, please install packmol on your machine and add the following line to your bash profile.

export PACKMOL_PATH="path_to_packmol_executable_here"

Using MPmorph

Before diving headfirst into mpmorph, one should get familiar with how to run fireworks before jumping into mpmorph. Relevant features of fireworks include setting up a fireworks database, adding workflows to the database, configuring a conda environment for running fireworks on a supercomputer, launching jobs to a workload manager (using qlaunch), and monitoring fireworks jobs. For a quick tutorial, check out beginner and intermediate atomate/fireworks tutorials from our 2019 Materials project workshop.

A sample of using mpmorph to run an AIMD simulation at 1500K for 200ps (100k steps at 2fs/step) is shown below:

from mpmorph.workflows.converge import get_converge_wf
from pymatgen.ext.matproj import MPRester

from fireworks import LaunchPad

mpr = MPRester()
structure = mpr.get_structure_by_material_id('mp-1143')
structure.make_supercell([3, 3, 3])

wf = get_converge_wf(structure, temperature = 750, target_steps = 100000)

lp = LaunchPad.auto_load()
lp.add_wf(wf)

To generate an amorphous structure, run the following code:

from mpmorph.runners.amorphous_maker import get_random_packed
from mpmorph.workflows.converge import get_converge_wf
from fireworks import LaunchPad

structure = get_random_packed('Li', target_atoms=100)

wf = get_converge_wf(structure, temperature = 5000, target_steps = 10000)

lp = LaunchPad.auto_load()
lp.add_wf(wf)

Customizing runs

At the simplest level of customizing the workflow, one can change the temperature, total number of steps by changing the args passed to get_converge_wf().

For more advanced changes, pass a dictionary for "converge_args", "spawner_args", or "prod_args". This allows customization of vasp inputs, starting and ending temperatures, number of steps, and convergence crieteria.

Examples:

Changing the rescale parameter, to make for larger volume changes at each stage:

from mpmorph.runners.amorphous_maker import get_random_packed
from mpmorph.workflows.converge import get_converge_wf
from fireworks import LaunchPad

structure = get_random_packed('Li', target_atoms=100)

spawner_args = {'rescale_params': {'beta': 5e-6}}
wf = get_converge_wf(structure, temperature = 5000, target_steps = 10000, spawner_args=spawner_args)

lp = LaunchPad.auto_load()
lp.add_wf(wf)

Changing precision of production runs to PREC=Low:

from mpmorph.runners.amorphous_maker import get_random_packed
from mpmorph.workflows.converge import get_converge_wf
from fireworks import LaunchPad

structure = get_random_packed('Li', target_atoms=100)

prod_args = {'optional_fw_params': {'override_default_vasp_params': {'user_incar_settings': {'PREC': 'Normal'}}}}
wf = get_converge_wf(structure, temperature = 5000, target_steps = 10000, prod_args = prod_args)

lp = LaunchPad.auto_load()
lp.add_wf(wf)

Citation

If you use mpmorph, please cite the following papers: