A dedicated Python script for making smeared C-K edge spectra from the hdf5 dataset of eigenvalues and dynamical structure factors
- Some functions for making Gaussiean smeared spectra database from the hdf5 dataset of eigenvalues and dynamical structure factors
- Commandline interface
ck_edge_maker
- Dataset class for Pytorch Geometric
ck_edge_maker.dataset.CK
The hdf5 dataset of eigenvalues and dynamical structure factors will be available along with a data paper
Clone this repository and run pip install
:
$ git clone https://github.com/nmdl-mizo/ck_edge_maker.git
$ cd ck_edge_maker
$ pip intall .
or directry run pip install
:
$ pip install git+https://github.com/nmdl-mizo/ck_edge_maker
If you use the dataset class for PyTorch Geometric, include extras_require pyg:
$ pip install "ck_edge_maker[pyg] @ git+https://github.com/nmdl-mizo/ck_edge_maker"
To uninstall, use pip:
$ pip uninstall ck-edge-maker
Calculate C-K edge spectra with Gaussian smearing from hdf5 spectral dataset
-
positional arguments:
{site,mol}
calculation task, site or molinput
path to the input database hdf5 fileoutput
path to the output hdf5 file
-
optional arguments:
-h, --help
show this help message and exit--sigma SIGMA
Gaussian smearing parameter in eV. default 0.3 eV--margin MARGIN
margin in eV for sampling energy range. ignored when both min and max is specified. default 3.0 eV--res RES
resolution in eV for sampling energy range, default 0.1 eV--min MIN
minimum energy in eV for sampling energy range--max MAX
maximum energy in eV for sampling energy range--sum
calculate sum of dynamic structure factors for molecular spectra if specified. calculate weighted average if not specified. only valid when task = mol--nototal
do not include total spectra if specified-f, --forceoverwrite
Overwrite output file if specified
First, prepare the hdf5 dataset of eigenvalues and dynamical structure factors. The following is an example where the dataset is stored in the current directory with the name "site_eigen_dsf.hdf5".
# Create a hdf5 of molecular spectral dataset named "mol_spectra_0.5eV.hdf5" with 0.5eV Gaussian smearing, 0.1eV sampling step, and 5eV margin
ck_edge_maker mol site_eigen_dsf.hdf5 mol_spectra_0.5eV.hdf5 -f --sigma 0.5 --res 0.1 --margin 5
# Create a hdf5 of site specific spectral dataset named "site_spectra_0.5eV.hdf5" with 0.5eV Gaussian smearing, 0.1eV sampling step, and 5eV margin
ck_edge_maker site site_eigen_dsf.hdf5 site_spectra_0.5eV.hdf5 -f --sigma 0.5 --res 0.1 --margin 5
The dataset class for Pytorch Geometric is also available from ver.1.1.0.
- Install ck_edge_maker with pyg extras_require
pip install ck_edge_maker[pyg]
- Prepare site specific spectral dataset and place it in "dataset/raw" directory.
- Import CK class
from ck_edge_maker.dataset import CK dataset = CK( root="dataset", energies=(288, 310, 256), directional=True )
- python (=>3)
- h5py
- numpy
- tqdm
- scipy
- torch
- torch_geometric
- kiyou
- Mizoguchi Lab.
The source code is licensed MIT.