Graphics processing units accelerated semiclassical initial value representation molecular dynamics.

Area of Science:

• Computational Chemistry
• Molecular Spectroscopy
• Quantum Mechanics

Background:

• The Semiclassical Initial Value Representation (SC-IVR) is a powerful method for calculating vibrational molecular spectra.
• Traditional SC-IVR implementations can be computationally intensive, limiting their application to larger systems.
• Graphics Processing Units (GPUs) offer massive parallel processing capabilities that can accelerate such calculations.

Purpose of the Study:

• To implement and evaluate a Graphics Processing Units (GPUs) version of the Semiclassical Initial Value Representation (SC-IVR) propagator.
• To assess the performance and efficiency of the GPU-SC-IVR implementation for vibrational molecular spectroscopy.
• To demonstrate the environmental benefits of GPU acceleration in computational chemistry.

Main Methods:

• Implementation of the time-averaging formulation of SC-IVR for power spectrum calculations on GPUs.
• Utilizing NVIDIA Tesla C2075 and Kepler K20 GPUs for the computational tasks.
• Comparing the computational time and scaling against Intel Core i5 and Intel Xeon E5-2687W CPUs.

Main Results:

• The GPU-SC-IVR implementation achieved perfect agreement with benchmark vibrational frequencies for four molecules.
• Significant reductions in computational time and power consumption were observed using GPUs compared to CPUs.
• The study identified critical aspects and challenges related to the GPU implementation of semiclassical methods.

Conclusions:

• GPU acceleration provides a substantial speed-up for SC-IVR calculations in vibrational molecular spectroscopy.
• The developed GPU-SC-IVR method is computationally efficient and demonstrates significant power savings.
• This approach offers a more environmentally friendly alternative for complex molecular simulations.