Symmetrisation schemes for global optimisation of atomic clusters

Area of Science:

• Computational Chemistry
• Materials Science
• Optimization Algorithms

Background:

• Identifying global minima in atomic and molecular clusters is a complex optimization challenge.
• High-symmetry structures often exhibit extreme energy values, influencing optimization strategies.

Purpose of the Study:

• To benchmark optimization procedures that leverage approximate symmetry for locating global minima.
• To assess the efficiency gains of symmetry-exploiting algorithms in cluster energy landscape exploration.

Main Methods:

• Theoretical analysis to understand the role of approximate point group symmetry in energy landscapes.
• Implementation of symmetry-exploiting algorithms within the GMIN program.
• Combination of approximate symmetry methods with basin-hopping and genetic algorithms.

Main Results:

• Benchmarks conducted on 38-, 75-, and 98-atom Lennard-Jones (LJ) clusters, which are multiple-funnel systems.
• Significant reduction (up to two orders of magnitude) in mean time to locate the global minimum.
• Moderate efficiency improvements observed for simpler single-funnel systems like LJ(55) and LJ(74).

Conclusions:

• Exploiting approximate symmetry is a highly effective strategy for accelerating global minimum searches in complex cluster systems.
• The developed algorithms offer substantial computational savings for atomic and molecular cluster optimization.
• Symmetry-based approaches enhance the efficiency of established optimization techniques like basin-hopping and genetic algorithms.