PDECO: parallel differential evolution for clusters optimization.

Area of Science:

• Computational Chemistry
• Materials Science
• Optimization Algorithms

Background:

• Optimizing atomic and molecular clusters with many atoms is computationally intensive and prone to local optima.
• Existing methods struggle with the scale and complexity of large cluster optimization problems.

Purpose of the Study:

• To develop and evaluate a parallel differential evolution (DE) optimization scheme for large-scale atomic clusters.
• To enhance the global searching ability and computational efficiency of cluster optimization.

Main Methods:

• A modified DE algorithm incorporating improved genetic operators.
• A parallel strategy with a migration operator to handle large computational demands and local optima.
• Application to Lennard-Jones (LJ) clusters, Gupta-potential Cobalt (Co) clusters, and Gupta-potential Platinum (Pt) clusters.

Main Results:

• The parallel DE algorithm demonstrated superior performance over previous methods in terms of success rate, convergence speed, and global searching ability for LJ and Co clusters.
• Significant enhancement in performance for large or challenging LJ clusters.
• Identified magic numbers (13, 38, 54, 75, 108, 125 atoms) for Pt clusters, showing a correlation between stability and a larger energy gap (HOMO-LUMO).
• Pt38 is predicted to be more catalytically active than Pt75.

Conclusions:

• The proposed parallel DE optimization scheme is effective for large-scale atomic cluster optimization.
• Stable Pt clusters exhibit larger energy gaps, suggesting potential for tailored electronic properties.
• The findings provide insights into the catalytic activity of Pt clusters of varying sizes.