Particle-swarm structure prediction on clusters.

Area of Science:

• Computational Chemistry
• Materials Science
• Chemical Physics

Background:

• Predicting the lowest-energy structures of atomic clusters is crucial for understanding their properties.
• Traditional methods often struggle with the vast search space and complexity of potential energy surfaces.

Purpose of the Study:

• To develop an efficient and reliable computational method for atomic cluster structure prediction.
• To improve the exploration of potential energy surfaces for non-periodic systems.

Main Methods:

• Generalization of particle swarm optimization (PSO) with a local search capability.
• Development and application of a bond characterization matrix (BCM) for structural similarity assessment and search space definition.
• Integration of point group symmetries and the Metropolis criterion to enhance structural diversity and convergence to low-energy states.

Main Results:

• The developed method demonstrates high search efficiency, validated on Lennard-Jones clusters up to 150 atoms.
• Successful application to predict novel structures for medium-sized lithium (Li(n)) clusters (n = 20, 40, 58).
• The BCM technique effectively prunes redundant structures and refines search spaces.

Conclusions:

• The new PSO-based methodology offers a reliable and efficient approach for atomic cluster structure prediction.
• The integration of symmetry and advanced search criteria significantly improves the handling of large and complex systems.
• This method shows promise as a leading tool for computational cluster studies.