Benchmarking non-adiabatic quantum dynamics using the molecular Tully models.

Area of Science:

• * Computational Chemistry
• * Quantum Dynamics
• * Spectroscopy

Background:

• * On-the-fly non-adiabatic dynamics methods are crucial for characterizing system evolution post-light absorption.
• * These methods bypass the need for pre-computed potential energy surfaces, calculating key properties dynamically.
• * Benchmarking approximate methods against established models is essential for accuracy assessment.

Purpose of the Study:

• * To validate the on-the-fly direct dynamics variational multi-configuration Gaussian (DD-vMCG) method.
• * To compare DD-vMCG performance against established methods like Tully Surface Hopping (TSH) and Ab Initio Multiple Spawning (AIMS).
• * To utilize the Ibele-Curchod models (ethene, DMABN, fulvene) for benchmarking non-adiabatic behavior.

Main Methods:

• * Employed the direct dynamics variational multi-configuration Gaussian (DD-vMCG) method.
• * Utilized ethene, DMABN, and fulvene as molecular test systems (Ibele-Curchod models).
• * Compared on-the-fly DD-vMCG results with parameterized linear vibronic potentials and other dynamics methods (TSH, AIMS).

Main Results:

• * DD-vMCG method successfully simulated non-adiabatic dynamics for the chosen molecular systems.
• * Crucial differences were observed between DD-vMCG, AIMS, and TSH in specific cases.
• * The classical nature and initial conditions of TSH simulations were identified as reasons for discrepancies.

Conclusions:

• * The DD-vMCG method provides a valuable tool for studying non-adiabatic dynamics.
• * Discrepancies highlight the importance of method choice and initial conditions in non-adiabatic simulations.
• * Further benchmarking is necessary to fully understand the accuracy and performance of various on-the-fly methods.