Artificial-Intelligence-Enhanced On-the-Fly Simulation of Nonlinear Time-Resolved Spectra

Area of Science:

• Physical Chemistry
• Computational Chemistry
• Spectroscopy

Background:

• Time-resolved spectroscopy reveals molecular structural dynamics crucial for biology and technology.
• Interpreting nonlinear femtosecond signals requires theoretical simulations, which are computationally expensive.
• High computational cost limits the use of first-principles methods for simulating transient absorption and 2D electronic spectra.

Purpose of the Study:

• To develop an artificial intelligence-enhanced protocol to reduce the computational cost of simulating nonlinear time-resolved electronic spectra.
• To make accurate simulations affordable for increasingly large polyatomic molecules.
• To enable wider adoption of advanced spectroscopic simulation techniques.

Main Methods:

• Developed an AI-enhanced protocol based on the doorway-window approach.
• Utilized on-the-fly surface-hopping simulations.
• Validated the protocol using the pyrazine molecule as a prototypical system.

Main Results:

• Achieved a drastic reduction in computational cost, cutting expenses by at least 95% compared to pure first-principles simulations.
• Produced highly precise spectra for pyrazine, demonstrating high accuracy relative to ab initio references.
• Showcased the protocol's applicability for polyatomic molecules of increasing size.

Conclusions:

• The AI-enhanced protocol significantly lowers the computational barrier for simulating nonlinear time-resolved electronic spectra.
• This method provides a cost-effective and accurate approach for molecular dynamics studies.
• The protocol facilitates the detailed investigation of molecular processes in both biological and technological fields.