A simple graphics processing unit-accelerated propagation routine for laser pulses in the strong-field regime

Area of Science:

• Computational physics
• Nonlinear optics
• Laser-matter interactions

Background:

• Simulating ultrashort and intense laser pulse propagation is crucial for understanding laser-matter interactions.
• Existing numerical methods can be computationally intensive, limiting research scope.
• Efficient simulation tools are needed to advance fields like plasma physics and materials science.

Purpose of the Study:

• To present a simple, GPU-accelerated numerical routine for simulating laser pulse propagation.
• To demonstrate the implementation of a frequency-domain solver using an extended Crank-Nicolson algorithm on Nvidia GPUs.
• To achieve significant speed-up compared to traditional CPU-based methods.

Main Methods:

• Developed a Graphics Processing Unit (GPU)-accelerated routine using Nvidia CUDA C++.
• Implemented a frequency-domain solution of Maxwell's wave equation with an extended Crank-Nicolson algorithm.
• Simulated the propagation of a near-infrared laser pulse through a partially ionized atomic gas.

Main Results:

• The GPU-accelerated routine achieves a significant speed-up factor.
• The method is easy to implement with basic CUDA and C++ knowledge.
• Simulations showed the GPU version is up to 198 times faster than a comparable CPU implementation.

Conclusions:

• The presented GPU-accelerated method offers an efficient and accessible tool for simulating laser pulse propagation.
• This approach can accelerate research in nonlinear optics and laser-plasma interactions.
• The ease of implementation lowers the barrier for researchers to utilize GPU acceleration.