GPU-accelerated on-the-fly nonadiabatic semiclassical dynamics

Area of Science:

• Computational Chemistry
• Quantum Dynamics
• Materials Science

Background:

• Simulating nonadiabatic excited-state dynamics is crucial for understanding chemical reactions and material properties.
• Existing methods often face computational limitations for complex systems.

Purpose of the Study:

• To introduce PySCES, a novel computational tool for GPU-accelerated on-the-fly nonadiabatic dynamics.
• To demonstrate the efficiency and capabilities of PySCES for complex molecular systems.

Main Methods:

• Interfacing the linearized semiclassical dynamics approach with the TeraChem electronic structure program.
• Developing a Python code (PySCES) for semiclassical dynamics with on-the-fly electronic structure calculations.
• Implementing parallelization across multiple GPU nodes for enhanced performance.

Main Results:

• Successfully demonstrated the computational workflow of PySCES.
• Presented timing benchmarks for two systems: fulvene in acetonitrile and a zinc-phthalocyanine/fullerene charge transfer system.
• Showcased the ability to model ultrafast charge transfer dynamics across multiple electronic states.

Conclusions:

• PySCES provides an efficient semiclassical approach for modeling nonadiabatic excited-state dynamics.
• This implementation facilitates the study of complex molecules, materials, and condensed-phase systems.
• The GPU acceleration significantly enhances the computational feasibility of these simulations.