Accelerating Anharmonic Spectroscopy Simulations via Local-Mode, Multilevel Methods.

Area of Science:

• Computational Chemistry
• Theoretical Spectroscopy
• Quantum Mechanics

Background:

• Anharmonic vibrational spectra simulations offer deep insights into molecular behavior.
• Generating mode-coupling potentials is a major computational challenge for *ab initio* simulations.
• Accurate prediction of molecular vibrational spectra is crucial for understanding chemical reactions and properties.

Purpose of the Study:

• To develop and validate an efficient computational strategy for *ab initio* anharmonic vibrational spectra.
• To reduce the computational bottleneck associated with generating mode-coupling potentials.
• To achieve sub-wavenumber accuracy in vibrational spectra with significantly reduced computational cost.

Main Methods:

• Combines a local-mode representation with multilevel computational methods.
• Employs a low-level quantum chemistry method for pre-screening pairwise couplings.
• Utilizes distance-based truncation of low-level predictions to further optimize efficiency.

Main Results:

• Achieved sub-wavenumber fidelity with reference vibrational transitions using a small fraction of target-level couplings.
• Demonstrated a 5-fold reduction in computational cost for alanine dipeptide VSCF spectra.
• The overhead of predicting couplings with the new method is minimal.

Conclusions:

• The combined local-mode and multilevel approach offers a computationally efficient route to accurate anharmonic vibrational spectra.
• This method significantly accelerates simulations, especially for larger molecular systems.
• The approach shows considerable promise for future advancements in computational spectroscopy.