GPU accelerated Monte Carlo simulation of pulsed-field gradient NMR experiments

Area of Science:

• Computational Physics
• Magnetic Resonance Imaging
• Materials Science

Background:

• Monte Carlo (MC) simulations are crucial for Nuclear Magnetic Resonance (NMR) diffusion measurements in porous media.
• Current simulations require extensive computation time (hundreds of milliseconds) and numerous trajectories for statistical convergence.

Purpose of the Study:

• To accelerate MC simulations of diffusion for NMR measurements.
• To enable advanced analysis in scenarios where analytical models are insufficient.

Main Methods:

• Parallelization of MC diffusion simulation code for execution on graphics processing units (GPUs).
• Leveraging the parallel processing power of GPUs to significantly reduce computation time.

Main Results:

• Achieved acceleration of MC diffusion simulations by over three orders of magnitude (1000x).
• Demonstrated the feasibility of GPU acceleration for complex diffusion modeling.

Conclusions:

• GPU-accelerated MC simulations open new possibilities for experimental design and data analysis in NMR.
• This approach is broadly applicable to complex geometries, non-ideal gradient pulse conditions, and diffusion-weighted MRI of tissues like the lungs and brain.