pyMSFT 1.0.2

python MultiSlice Fourier-Transform toolbox

  1.0.2      ___  ___ ___________ _____
  CC-BY-4.0  |  \/  |/  ___|  ___|_   _|
  _ __  _   _| .  . |\ `--.| |_    | |
 | '_ \| | | | |\/| | `--. \  _|   | |
 | |_) | |_| | |  | |/\__/ / |     | |
 | .__/ \__, \_|  |_/\____/\_|     \_/
 | |     __/ |        python Multi-Slice
 |_|    |___/         Fourier-Transforms

pyMSFT is a python toolbox for accurately and efficiently simulating wide-angle coherent diffractive imaging. For details on the physics implemented in this code, please refer to the following publication: https://doi.org/10.1002/lpor.202502001

Installation

We strongly recommend setting up a virtual environment for the installation, i.e. using conda. This code has been tested for python 3.12.

conda create -n pymsft python=3.12
conda activate pymsft

At this point you can install the package using pip with different options:

• To install the basic package without GPU support and Mie scattering support, simply run:

  pip install pyMSFT
  

• To take advantage of GPU acceleration we are using CuPy. See the installation instructions for more details. The simplest way to install it is currently via conda:

  conda install -c conda-forge cupy==13.6.0
  

  Note that an NVIDIA GPU with the corresponding driver installed is required.
• To calculate the Mie solutions for spherical particles you must also install corresponding Mie solver (we are using scattnlay). This can be installed with:

  pip install pyMSFT[mie]
  

Usage

A simple example for a multislice simulation is shown below. Note that most of these examples require matplotlib for visualization to be installed.

import numpy as np
from pyMSFT.shapes import Spindle
from pyMSFT.simulation import PMSFTSimulation

import matplotlib.pyplot as plt
from matplotlib.colors import LogNorm

wavelength = 13.5e-9            # wavelength of the light in meters
Rsphere = 200e-9                # radius of the sphere in meters

refIndex = 0.89 + 1j * 0.09     # complex refractive index of Silver at 13.5 nm (92 eV)

npix_fft = 256  # number of pixels in the FFT (should be a power of 2)
npix_real = 64  # number of pixels in the real space simulation box - this will be zero padded to num_pix_fft internally

# setup the geometry and simulation
geometry = Spindle(refractive_index=refIndex, npix=npix_real, x_rotation=45 * np.pi / 180, y_rotation=30 * np.pi / 180, z_rotation=0)
simulation = PMSFTSimulation(npix_fft=npix_fft, npix_real=npix_real, wavelength=wavelength, box_size=2 * Rsphere, verbose=True)

# run the simulation and get the diffraction image
diff_image, (QX, QY) = simulation.get_diffraction_image(geometry, output_axes=True)

# visualize the result
_, axs = plt.subplots(1, 1, figsize=(6, 4))
q_extent = QX[0, 0], QX[-1, 0], QY[0, 0], QY[0, -1]
barn = 10 ** -28            # barn unit in m^2
diff_image /= barn ** 2     # convert to barn^2

img = axs.imshow(diff_image, extent=q_extent, cmap='turbo', norm=LogNorm(vmin=1e24, vmax=1e28))
plt.colorbar(img)
axs.set_title(r'differential cross section on Ewald sphere in [b$^2$]')

axs.set_xlabel('QX [1/m]')
axs.set_ylabel('QY [1/m]')
axs.set_aspect('equal')
plt.show()

This will create a diffraction pattern of a silver spindle illuminated with 13.5 nm light and should look like this:

More examples can be found in the examples folder, e.g. for:

• Comparing different simulation methods with an exact Mie solution for spheres: examples/compare_simulation_methods.py

And more are planned and coming soon, e.g. for:

• Visualizing the experimental setup regarding the internal orientations and axes: examples/visualize_setup.py
• Comparing the performance of CPU and GPU simulations: examples/compare_cpu_gpu.py
• Compare different detector geometries: examples/compare_detectors.py
• Showcase the different available sample shapes: examples/showcase_shapes.py

Citation

If you use this code for your research, please cite it as:

P. Tuemmler, J. Apportin, T. Fennel, and C. Peltz, “Fast Simulation of Wide-Angle Coherent Diffractive Imaging.” Laser Photonics Rev (2025): e02001. https://doi.org/10.1002/lpor.202502001