High Entropy Alloys

Selected Python codes for High Entropy Alloys analysis and tools

Project maintained by Asif-Iqbal-Bhatti Hosted on GitHub Pages — Theme by mattgraham

High-Entropy-Alloys — Computational Toolkit

A collection of Python (and some C++/Fortran) scripts for first-principles and machine-learning-based analysis of High-Entropy Alloys (HEAs). The tools cover the full workflow from structure generation and thermodynamics through defect mechanics, machine-learned potentials, and deep-learning property prediction.

Scripts accompanying the peer-reviewed publication:

doi.org/10.1016/j.commatsci.2024.113196

Repository layout

High-Entropy-Alloys/
├── src/                              # Core structure and thermodynamics scripts
├── MLIP_MonteCarlo/                  # Monte Carlo with machine-learned potentials
├── VASP_MLFF/                        # VASP machine-learned force-field utilities
├── VASP-MDML-DataExtraction/         # VASP → Extended-XYZ / ML-dataset preparation
├── MaterialsPropertyAnalysis_ML_DNN/ # ML / deep-learning property prediction
├── Elastic_tensor_from_Lammps/       # Elastic tensor extraction via LAMMPS
├── surface_states_alloys/            # Surface energy calculations
└── utilities/                        # Plotting, formation energy, SRO, dislocations, …

Modules

src/ — Structure generation & core thermodynamics

Script Purpose
SQS_structure_gen.py Generates Special Quasi-random Structures (SQS) for multi-component HEAs
mcsqs2poscar.py / sqs2poscar_modified_AIB.cpp Converts MCSQS output to VASP POSCAR format
get_mcsqs_lattices.py Prepares BCC/FCC/HCP lattice inputs for MCSQS
PMSROCalculatorPyScale2.py Computes PM-SRO and WC-SRO short-range order parameters across neighbour shells
get_ConfigurationEntropy_HEA.py Calculates configurational entropy (ΔS_conf = −k_B Σ c_i ln c_i) from CIF files, with temperature sweep and CSV export
GSFE_BCC.py / GSFE.sh Generalized Stacking Fault Energy (GSFE) calculations for BCC crystals
BCCScrew_dislocation.py / BCCScrew_dislocation.f90 Screw dislocation core analysis for BCC lattices
lattice_mismatch_measure.py Quantifies lattice mismatch between alloy components
scalingPOSCAR.py / supercell_rand.py POSCAR scaling and random supercell construction
set_orderTypes_POSCAR.py / set_shake_atom_lattice.py Atomic ordering and thermal displacement tools
read_POSCAR.cpp Lightweight C++ POSCAR reader

MLIP_MonteCarlo/ — Monte Carlo with machine-learned potentials

MLIP_MonteCarlo_HEA.py — Metropolis Monte Carlo driven by a GRACE machine-learned interatomic potential. Performs random atomic substitutions on a supercell, accepts/rejects moves based on energy at a target temperature, and saves the lowest-energy configuration found.

VASP_MLFF/ — VASP machine-learned force-field utilities

Script Purpose
get_data.py Extracts training frames from VASP ML-FF runs
get_MLData_py4vasp.py Reads VASP output via py4vasp for ML dataset construction
get_VASP_ML_ERR_Analysis.py Analyses force/energy errors of the VASP ML-FF against DFT reference

VASP-MDML-DataExtraction/ — VASP → ML-ready datasets

Script Purpose
vaspouth5_to_extxyz.py Converts vaspout.h5 trajectories to Extended XYZ
get_MLAIMDVASP_to_EXTXYZ.py Extracts AIMD frames from VASP output to Extended XYZ
1_MPtrj_Dataset_Query_and_Parsing_Criteria.py Queries and filters the MPtrj dataset (energy convergence, GGA/GGA+U matching)
get_PureElementEnergyRef.py Retrieves DFT reference energies for pure elements
generate_PESPlot_from_DFTData.py / get_PESPlot_DFTData.py Potential energy surface visualisation via UMAP dimensionality reduction
get_plot_MSD_GPU.py / plot_MSD_from_VASPKIT.py GPU-accelerated (CuPy) and standard MSD analysis
get_struct_from_Alexandria.py Downloads structures from the Alexandria database
test_model_from_SevenNet.py Evaluates a SevenNet MLIP against DFT data

MaterialsPropertyAnalysis_ML_DNN/ — Machine learning & deep learning

Script Purpose
ML_DNN_with_TensorFlowKeras/DeepNeuralNetwrok_TF_HEA.py Deep neural network (TF/Keras) for HEA property regression
ML_DNN_with_TensorFlowKeras/LinearRegression_TF_HEA.py TensorFlow linear regression baseline
ML_DNN_with_TensorFlowKeras/get_gaussianProcessRegression.py Gaussian process regression for property prediction
pred_short_range_order_pytorch.py PyTorch neural network trained on Materials Project data to predict SRO / alloy properties

Elastic_tensor_from_Lammps/

elastic_tensor_lammps.py — Automates extraction and symmetrisation of the full elastic stiffness tensor from LAMMPS deformation runs.

surface_states_alloys/

Script Purpose
1_generate_surface_pmg.py Generates surface slabs using Pymatgen
get_surface_energy_using_pymatgen.py Computes surface energy (J/m²) from VASP CONTCAR/OUTCAR files relative to a bulk reference

utilities/ — Supporting tools

Sub-directory / script Purpose
HEA_composition.py Composition parsing and weight-fraction ↔ atom-fraction conversion
configuration_entropy.py Standalone configurational entropy calculator
coordination_analysis.py Coordination number analysis from trajectory data
Poscar_colorPlotter.py Colour-coded POSCAR structure visualisation
get_InitialRelax_from_ASE.py Initial structure relaxation via ASE
get_dispersion.py Phonon/electronic dispersion utilities
ClusterExpansion_tools/icet_ClusterExpansion.py Cluster expansion with ICET
ConversionData_tools/ Unit/format converters (DAT → JSON, wt% → at%)
DislocationDipole_tools/ Dislocation dipole elastic energy, Frenkel-Kontorova model, core energy
FormEnergy_tools/calculate_formation_energy.py Formation energy calculation from DFT total energies
MCSQS_tools/ Parallel SQS generation, SRO via pyscal, MCSQS file numbering
MonteCarlo_tools/Metropolis_Monte_carlo.py Metropolis Monte Carlo reference implementation
PhaseDiagram_tools/phase_diagram_for_HEA.py HEA phase diagram construction
plot_tools_HEA/ Vitek maps, box plots, bar charts, spider plots, 3-D spline fitting, PDOS plotting, Fresnel rendering

Dependencies

Category Packages
Core scientific numpy, scipy, matplotlib, pandas
Atomistic ase, pymatgen, pyscal
VASP interface py4vasp
Machine learning scikit-learn, tensorflow / keras, torch
MLIPs grace (GRACE potential), sevenn (SevenNet)
Dimensionality reduction umap-learn, plotly
GPU acceleration cupy (optional, for MSD on CUDA hardware)
Other icet (cluster expansion), atomsk (structure generation, external binary)

Python 3.8+ is required throughout.

Citation

If you use these scripts in your research, please cite:

Bhatti, A. I. et al., Computational Materials Science, 2024. https://doi.org/10.1016/j.commatsci.2024.113196

License

Distributed under the GNU General Public License v3.0. See LICENCE for details.

Disclaimer: Scripts are provided as-is with no warranty. Please review each script carefully before use in a research workflow.