Basic python code for phasing three-dimensional diffraction volumes

To install run:

$ git clone https://github.com/andyofmelbourne/3D-phasing.git && cd 3D-phasing && pip install -e .

To install dependencies through conda, run:

$ conda env create -f environment.yml -n phasing
$ conda activate phasing

A basic example with a 3D duck:

$ phasing_init.py && cd example
$ python reconstruct.py config.ini

See also:

• config_background.ini: radial background retrieval, and
• config_voxel_number_support.ini: unknown support uses the number of voxels in the sample
• config_repeats.ini: merge many independent recontructions with unkown support and background retrieval

For this last example:

$ python reconstruct.py config_repeats.ini -i     
$ mpirun -np 20 python phase.py input.h5     

The first line just makes the input file while the second runs the script with 20 cpu cores, each core repeating the reconstructions 10 times for a grand total of 200 recontructions.

When complete, you may display the output:

$ python display_phasing.py output.h5 output

Of course the above is all just padding around:

import phasing
import numpy as np
O = np.random.random((64,64,64))
S = np.zeros(O.shape, dtype=np.bool)
S[:16,:16,:16] = True
O *= S
I = np.abs(np.fft.fftn(O))**2
Oout, info = phasing_3d.DM(I, 100, support = S)
Oout, info = phasing_3d.ERA(I, 100, support = S, O = Oout)

Try it yourself, just copy the above text then:

$ ipython
>>> %paste

and watch the magic happen!