For a thorough description of the project, please see the paper by Özer et al.: https://doi.org/10.48550/arXiv.2204.00434.
The following guidelines assume that the user runs a conda distribution, i.e. Anaconda or Miniconda.
- Create/activate new conda env by running:
conda create -n pydatarecognition python=3 conda activate pydatarecognition
- Navigate to the main pydatarecognition directory and run:
conda install --file requirements/run.txt pip install -r requirements/pip_requirements.txt
- Install the package by navigating to the main pydatarecognition
directory and run:
python setup.py install
Currently, the program should be run from a directory containing the cif files.
Within docs/examples
, example cifs are located in the cifs/measured
and cifs/calculated
subdirectory, e.g., in docs/examples/cifs/measured
.
Within docs/examples/powder_data
, three examples on input data files are available:
- 01_Mg-free-whitlockite_wl=1.540598.txt
- 02_BaTiO3_wl=0.1665.txt
- 03_KNaLi-NbMnO3_perovskite_wl=1.5482.txt
With your pydatarecognition
conda env activated, to get information on how to run the program type:
pydr --help
or
pydatarecognition -h
The program expects a syntax somewhat similar to:
python pydatarecognition.main -i INPUTFILE --xquantity XQUANTITY --xunit XUNIT -w WAVELENGTH
For a full description, please run the program with the help flag as shown above.
Navigate to docs/examples/cifs/measured
.
pydr -i ../../powder_data/01_Mg-free-whitlockite_wl=1.540598.txt --xquantity twotheta --xunit deg -w 1.540598
pydr -i ../../powder_data/02_BaTiO3_wl=0.1665.txt --xquantity twotheta --xunit deg -w 0.1665
pydr -i ../../powder_data/03_KNaLi-NbMnO3_perovskite_wl=1.5482.txt --xquantity twotheta --xunit deg -w 1.5482
Navigate to docs/examples/cifs/calculated
and rerun the commands above.
Output files will be available in the _output
folder created in the current working directory, i.e.
docs/examples/_output
.