maxiv-science / bcdi-assemble Goto Github PK

This is a utility library which enables the assembly of Bragg Coherent Diffraction Imaging (BCDI) frames of unknown orientations, such as collected via spontaneous rocking curves. The library is based on a Python module bcdiass which provides the basic operators of the algorithm as well as utilities for both simulation of experimental data and assembly of the same.

The core assembly library depends on standard python components such as numpy, scipy, matplotlib, scikit-image. In addition, simulating datasets from geometrical model particles currently depends on the 3dBPP branch of ptypy (registration needed) as well as the NanoMAX beamline utitility library.

The library installed with setuptools, for example like

cd bcdi-assembly
python setup.py --user

Assembly procedures are easily scripted with the utilities provided. Examples are provided in the examples folder, and a minimal example might look as follows.

import numpy as np
from bcdiass.utils import *

# input and parameters
data = np.load('data.npz')['frames']
Nj, Nl = 25, 25

# initial model
envelope = generate_envelope(Nj, data.shape[-1], support=.2)
W = generate_initial(data, Nj)

# assembly loop
for i in range(100):
    W, Pjlk, timing = M(W, data, Nl=Nl, beta=1e-4)
    W, error = C(W, envelope)