Ultrafast electron diffraction pattern simulations using GPU technology. Applications to lattice vibrations.

Area of Science:

• Materials Science
• Computational Physics
• Crystallography

Background:

• Electron diffraction is a key technique for materials characterization.
• Simulating electron diffraction patterns, particularly using the multislice method, is computationally intensive.
• Accelerating these simulations is crucial for efficient materials analysis.

Purpose of the Study:

• To develop and implement a GPU-accelerated method for electron diffraction pattern simulations.
• To enhance the speed of multislice calculations through optimized phase grating generation.
• To demonstrate the method's effectiveness for large supercell arrays and thermal diffuse scattering simulations.

Main Methods:

• Implementation of a novel GPU-based algorithm for phase grating generation in multislice calculations.
• Utilizing graphical processing units (GPUs) to parallelize computationally demanding parts of the simulation.
• Encoding transmission functions representing atomic potentials for efficient calculation of thermal diffuse scattering.

Main Results:

• Significant speed-up of electron diffraction pattern simulations using the developed GPU method.
• Demonstrated effectiveness for large supercell arrays, a common requirement in advanced materials studies.
• Successful simulation of thermal diffuse scattering in silicon, highlighting the method's capability for analyzing vibrational modes.

Conclusions:

• GPU acceleration offers a cost-effective and powerful solution for speeding up electron diffraction simulations.
• The novel phase grating generation method is highly effective for accelerating multislice calculations.
• This approach facilitates more rapid and detailed analysis of materials structure and dynamics.