qemb 0.1.3

A script to do embedded cluster cutting and generate input files for calculations.

神秘的东方文字版

qemb

A toolscript to cut cluster for quantum embedding. Accept VASP structure (CONTCAR).

1. General

1.1 Installation

This project is now on PYPI, so a convenient way to install is like:

pip3 install qemb

or from git repo to get nightly update:

git clone https://github.com/Makarov3821/qemb
# if you are in China, gitee repo maybe faster
# git clone https://gitee.com/Marshall3821/qemb
cd qemb
pip3 install .

1.2 Basic usage

A help of input & output can be obtained by qemb -h:

$ qemb -h
usage: qemb [-h] --input INPUT_DIR --output OUTPUT_DIR -f CONTROL_FILE [-q QUEUE] [-p CORE_NUMBER] [--xc XC_FUNCTIONAL] [--basis BASIS_SET]

An embedding calculation tool from GGA@PBC to embedded hybrids

options:
  -h, --help           show this help message and exit
  --input INPUT_DIR    Input directory location
  --output OUTPUT_DIR  Output directory location
  -f CONTROL_FILE      control file location, must set
  -q QUEUE             Queue names. 'xp72mc10' is default.
  -p CORE_NUMBER       Core number used per node: 1/2/3/4/8/28...; 36 is default.
  --xc XC_FUNCTIONAL   Set xc if REST input generation is needed; b3lyp is default.
  --basis BASIS_SET    Set basis if REST input generation is needed; def2-SVP is default.

in which input_dir, output_dir, and control_file is a required field. We'll explain this three part in the following section.

Argument such as core_number, xc, is needed while using input file generation.

IMPORTANT NOTICE: This project is not intended for a pythonic package, so import qemb in Python will be void, only main program "qemb" is supported. This feature may be considered later in development.

2. Inputs and Outputs

2.1 Input_dir

For a classic catalytic procedure, you have to calculate both clean slab and absorbant on slab. So we suggest your folder is organized like:

root_dir/
├── absorb
│   ├── CO-bridge
│   │   └── VASP_CALCULATION_FILES
│   ├── CO-hollow
│   │   └── VASP_CALCULATION_FILES
│   └── CO-top
│       └── VASP_CALCULATION_FILES
└── clean
    └── VASP_CALCULATION_FILES

for best user experience, since some feature is upon this structure.

This script will obtain structure from CONTCAR, but POSCAR, POTCAR, KPOINTS, and a normal terminated OUTCAR will also be checked to assure it a true VASP running directory, not a dangling CONTCAR.

2.2 Output_dir

The script will find all possible dir from input dir, and reform them in a new output dir given. For example, if you use root_dir in 2.1 as your input_dir, your output_dir, maybe named root_dir_out, will be like:

root_dir_out/
├── absorb
│   ├── CO-bri
│   │   ├── ref
│   │   │   └── REGENERATED_FILES
│   │   └── run
│   │       └── INPUT_FILES
│   ├── CO-hollow
│   │   ├── ref
│   │   │   └── REGENERATED_FILES
│   │   └── run
│   │       └── INPUT_FILES
│   └── CO-top
│       ├── ref
│       │   └── REGENERATED_FILES
│       └── run
│           └── INPUT_FILES
└── clean
    ├── ref
    │   └── REGENERATED_FILES
    └── run
        └── INPUT_FILES

ref is designed to store original CONTCAR, and cluster generated. This design is meant to compatible for DFET method developed by Emily A. Carter, though this is not implemented at this time.

run is where to store input file of the cluster generated. Now we support REST program incar generation. More program support will be added in the future.

2.3 Control file

A full option template with detailed commemt is listed in emb.toml.template. We'll only explain modes of cutting in below.

2.3.1 Cutting_method

1: A sphere from the ori_point. Mostly hemisphere for surface slab.

2: A cylinder from the ori_point.

3: A near expanding. For surface like Cu(111) will be:

For surface like Mgo(100) will be:

6: Alternative way for NaCl-like system like:

2.3.2 Some example

A Cu31 clutser:

jobtype = 1.1
cluster_cutting_major_method = 2
cluster_cutting_minor_method = 5
cluster_layers = 2
cluster_expand_num = 2

A Mg-centered Mg9O9 cluster surrounded by a hemispherical pseudoes/point charges:

jobtype = 1.3
cluster_cutting_major_method = 2
cluster_cutting_minor_method = 7
cluster_layers = 2
cluster_expand_num = 1
charge_cutting = true
charge_cutting_sub_method = 1
charge_radius = 8.0

Another interesting thing is this script will automatically supercell when your input exceeded the original cell. Thus we can make:

A huge plain (based on a 4x4, 3 layered MgO(100) slab):

jobtype = 1.1
cluster_cutting_major_method = 2
cluster_cutting_minor_method = 7
cluster_layers = 2
cluster_expand_num = 15

Or a big rod (based on a 4x4, 3 layered MgO(100) slab):

jobtype = 1.1
cluster_cutting_major_method = 2
cluster_cutting_minor_method = 7
cluster_layers = 20
cluster_expand_num = 1