Geometry Optimization with Multilayer Methods Using Least-Squares Minimization.

Area of Science:

• Computational Chemistry
• Molecular Modeling
• Quantum Mechanics/Molecular Mechanics (QM/MM)

Background:

• Accurate molecular structure optimization is crucial in computational chemistry.
• Existing methods may face challenges with large systems or complex interactions.
• Multilayer QM/MM approaches offer a balance of accuracy and computational efficiency.

Purpose of the Study:

• To introduce a novel least-squares minimization scheme for optimizing molecular structures.
• To develop a mixed-coordinate optimization framework for QM/MM multilayer models.
• To enhance the efficiency and stability of molecular structure optimization.

Main Methods:

• Developed a mixed-coordinate optimization framework combining internal and Cartesian coordinates.
• Implemented a least-squares minimization method utilizing quasi-Newton steps for error evaluation.
• Employed mechanical embedding for inter-layer couplings and approximated the MM Hessian.

Main Results:

• The developed method demonstrated consistently stable and fast convergence.
• Successfully tested on both two-layer and three-layer QM/MM models.
• The mixed-coordinate approach effectively handles QM and MM regions.

Conclusions:

• The proposed least-squares minimization scheme is an effective tool for QM/MM molecular structure optimization.
• The method offers significant improvements in convergence speed and stability.
• This work advances the capabilities of multilayer QM/MM simulations.