Simulated electron diffraction data and MATLAB scripts for data simulation and sorting

There are four main MATLAB scripts, including:

main_generate_patterns_triclinic.m

This script generates a triclinic unit cell with a given volume and lattice parameters within specified ranges. It calculates 1,000 electron diffraction patterns, uniformly oriented in space, and produces black-and-white patterns in JPG format.

The key parameters include:

·        Electron wavelength (WL) in Å

·        Excitation error in Å⁻¹

·        Data resolution in Å⁻¹

These parameters are embedded in the file names. For example, the pattern:

DIFF_Fibonacci_CELL_3.5784_10.5383_17.1131_116.8042_103.9746_106.8257__WL_0.0251_ExcErr_0.01_RES_1_ori_-0.25_-0.62647_-0.73826.jpg

was generated for a unit cell with the following lattice parameters:

·        a = 3.5784 Å

·        b = 10.5383 Å

·        c = 17.1131 Å

·        α = 116.8042°

·        β = 103.9746°

·        γ = 106.8257°

·        Electron wavelength = 0.0251 Å

·        Excitation error = 0.01 Å⁻¹

·        Data resolution = 1 Å⁻¹

·        Orientation vector of the pattern = [-0.25, -0.62647, -0.73826]

The coordinate system is defined with the z-axis along the incident electron beam and the x- and y-axes in the detector plane. For simplicity, the detector distance is set to 0.

The script main_generate_patterns_monoclinic.m generates similar patterns with a similar parameter set but using a monoclinic unit cell. In this case, α and γ are fixed to 90°. Additionally, the script produces extinctions along the b-axis, according to 21 screw axis extinctions, which are consistent with the monoclinic crystal system.

The script main_image_sorter.m allows for manual labelling of data for training.

The script main_sort_results.m processes a result .json file and automatically sorts the diffraction patterns into folders based on their predicted class and confidence level.

The data folder contains simulated diffraction patterns (binary jpg) sorted into classes.