clst_order

Summary

Post-processing tools for calculating short-range order (SRO) parameters in substitutional lattice systems. These tools can be used to calculate the Warren-Cowley pair paramters as well as the cluster order parameters defined in the COP paper. The occurrence of arbitrary multi-point SRO motifs are systematically quantified by direct calculation normalized cluster probabilities. The software is written in python/cython for front-end accessiblity, and is compatible with any crystal structure file compatible with ASE. This includes trajectory files from LAMMPS, RASPA, VASP, Quantum Espresso, etc. Trajectories from Monte Carlo simulations in cluster expansion software packages such as CASM and ICET can also be handled. If you are quantifying SRO in Monte Carlo simulations based on cluster expansions, see the ICET branch for a fast on-the-fly cluster order parameter calculator.

Installation

Add the software directory to your pythonpath then run the clean.sh script

./clean.sh 

cython (recommended)

Link the appropriate gcc compiler in your bin. For example in an anaconda bin directory (if using anaconda 3):

ln -s  x86_64-conda_cos6-linux-gnu-gcc gcc  

Add numpy libraries to the build. (can be found with np.file after importing numpy as np in a python script)

export
CFLAGS=-I/path/to/python/packages/site-packages/numpy/core/include

python setup.py build_ext --inplace

Python only setup

./python_only.sh 

Usage

Python software is currently set up for single sublattice 2D/quasi-2D surfaces as well as 3D single sublattice substitutional crystals within (m x n x l) supercells. Functionality is being added for multiple sublattices in 2D and 3D crystals, and for evaluation of correlation functions. A minimalistic example is given below for a 5-layer AB surface alloy system with ASE Atoms object inputs. It calculates the probability of an A-A-A-A cluster, which can then be divided by marginal probabilities to obtain cluster order parameters. Currently, one cluster probability with one occupation is measured per calculation. This will soon be changed to the probabilities of all occupations in a single cluster.

from clst_prob import *
from lib_basis import *
prim_vectors, p_sites, p_numbers, p_sublattices = from_ase( prim_atoms , zsubs )

p = prim_cell(prim_vectors, p_sites, p_numbers, zsubs = [[0,1,2,3,4]],
sublattices=p_sublattices)

vectors, sites, numbers, sublattices = from_ase(supercell_atoms, zsubs)

clst_verts =[ [0.  , 0.  , 0.66290], 
[0.33333333 , 0.66666667 , 0.58145],
[0.66666667 , 0.33333333 , 0.50000], 
[0.00000 , 0.00000 , 0.41855] ]

#define all cluster vectors at these vertices with 2 degrees of freedom
clst =  cluster(clst_verts,get_variables(2))

supercell = cluster_cell(vectors,sites,numbers,p)
supercell.add_cluster(clst)

print( y.cluster_avg(is_slab=True))

Detailed descriptions of variables are provided in the clst_prob.py script and example execution is provided in the examples folder. This includes an example for parallelized execution of the software for several large trajectory files.