Sequential and parallel molecular mechanics calculations

Area of Science:

• Computational chemistry
• Molecular modeling
• Algorithm development

Background:

• Accurate molecular structure prediction is crucial in computational chemistry.
• Existing optimization algorithms may have limitations in efficiency and applicability.

Purpose of the Study:

• To describe a gradient algorithm for computational optimization of molecular structures.
• To discuss the practical implementation and compromises of this algorithm in a Fortran program (VULCAN).

Main Methods:

• Development and description of a gradient-based computational optimization algorithm.
• Implementation of the algorithm in a Fortran program (VULCAN) for sequential and parallel computing.
• Analysis of algorithm properties, acceleration techniques, and potential issues.

Main Results:

• The article details the VULCAN program, an implementation of the gradient algorithm.
• Comparison of various acceleration techniques for gradient algorithms.
• Identification of potential challenges and "traps" in the practical application of the algorithm.

Conclusions:

• The gradient algorithm offers a method for computational molecular structure optimization.
• The VULCAN program provides a practical, albeit with inherent compromises, implementation for diverse computing environments.
• Understanding algorithm properties and potential pitfalls is key for effective use.