"On-the-fly" : How to reliably and automatically characterize and construct potential energy surfaces.

Area of Science:

• Computational Chemistry
• Theoretical Chemistry
• Chemical Physics

Background:

• Accurate potential energy surfaces (PES) are crucial for understanding chemical reactions and molecular properties.
• Existing methods for PES exploration can be computationally expensive and may struggle with high-dimensional systems.
• Identifying key features like transition states and reaction pathways on PES is essential for chemical dynamics.

Purpose of the Study:

• To retool a previously developed code for on-the-fly PES exploration and automatic PES function construction.
• To leverage the code's global PES mapping ability for efficient identification of relevant configuration space regions.
• To demonstrate the code's utility in accurately locating minima and transition states, and in generating new PES functions via interpolation.

Main Methods:

• The study utilizes a computational code designed for global PES mapping.
• The code generates uniformly spaced density functional theory (DFT) or ab initio points.
• These points are used for precise PES feature localization (minima, transition states) or automated PES function creation.

Main Results:

• The retooled code successfully performs on-the-fly PES explorations and automatic PES construction.
• The method demonstrates efficiency in globally mapping PES, even for large-dimensional systems.
• Proof-of-concept applications on water, methane, and methylene imine validate the code's capabilities.

Conclusions:

• The developed computational approach offers a reliable and efficient tool for exploring and constructing molecular potential energy surfaces.
• This method accelerates the determination of critical regions in configuration space, aiding chemical dynamics studies.
• The code's ability to handle large dimensionality and automate PES generation provides a significant advantage in computational chemistry research.