Efficient softest mode finding in transition states calculations.

Area of Science:

• Chemical Physics
• Computational Chemistry

Background:

• Transition states are crucial for understanding chemical reaction dynamics.
• Current methods for locating transition states often require Hessian matrix calculations or have unpredictable computational costs.

Purpose of the Study:

• To develop a more efficient and numerically stable algorithm for first-principle transition state location.
• To address the limitations of existing softest-mode following methods.

Main Methods:

• Introduction of the locally optimal search direction finding algorithm (LOR).
• LOR extends the conjugate gradient method without requiring additional force calculations.
• Demonstration of improved numerical stability through force translation.

Main Results:

• The LOR algorithm demonstrates significantly faster performance compared to the original dimer conjugate gradient method.
• Experiments on the Baker test system validate the efficiency of LOR.
• The proposed method enhances numerical stability.

Conclusions:

• The LOR algorithm offers a more efficient approach to transition state location in chemical physics.
• This method provides a valuable alternative for studying reaction dynamics.
• LOR improves upon existing techniques by enhancing speed and numerical stability.