mrbcdi

Differentiable, multi-reflection Bragg coherent diffraction imaging (BCDI) for lattice distortion fields in crystals

Directly reconstruct the electron density and the deviatoric lattice distortion in compact crystals from independent BCDI data sets measured at a synchrotron facility. Useful for imaging isolated dislocations such as edge and screw disloca Written in Python + Pytorch.

Reference

S. Maddali et al, A differentiable forward model for the concurrent, multi-peak Bragg coherent x-ray diffraction imaging problem, 2022.

Datasets and reconstructions are available here through the Git LFS system and citeable on Zenodo: Alternately, the datasets may be downloaded from here.

System requirements

Software

Python modules, as determined by pipreqs:

    h5py==2.10.0
    logzero==1.7.0
    matplotlib==3.3.4
    numpy==1.20.1
    pyfftw==0.13.0
    scipy==1.6.2
    torch==1.10.0	
    tqdm==4.62.3

Hardware

These are based on what was available to me at the time of creating this repo.

1. CPU: Intel Xeon CPU E5-2623 v4 @ 2.60GHz or better.
2. GPU: Nvidia GeForce GTX 1650 or better.

Tutorials

The notebooks folder contains complete reconstruction tutorials, including dataset descriptions.

Acknowledgements

1. Dr. Stephan Hruszkewycz (ANL)
2. Dr. Saugat Kandel (ANL)