Exploration versus Exploitation in Global Atomistic Structure Optimization.

Area of Science:

• Computational Chemistry and Materials Science
• Chemical Physics
• Solid State Physics

Background:

• Discovering novel compounds requires navigating complex molecular energy landscapes.
• High-dimensional systems necessitate efficient search strategies due to computational limitations.
• Global minimum searches are crucial for identifying stable structures.

Purpose of the Study:

• To develop and quantify a method for balancing exploration and exploitation in global minimum searches.
• To improve the efficiency of discovering low-energy structures in computational chemistry and materials science.
• To analyze global search behaviors for optimizing search strategies.

Main Methods:

• Integrating Bayesian optimization concepts within an evolutionary algorithm framework.
• Quantifying exploration (probing unknown regions) and exploitation (investigating known low-energy regions).
• Applying the method to molecular compounds and crystalline surface reconstruction.

Main Results:

• The developed method significantly accelerates the search for global minima structures.
• An optimal balance between exploration and exploitation enhances search speed for molecular systems.
• The approach shows effectiveness, albeit to a lesser extent, for crystalline surface reconstruction.

Conclusions:

• Balancing exploration and exploitation is critical for efficient global minimum searches.
• The proposed method offers a faster route to discovering novel compounds and materials.
• Further analysis provides grounds for optimizing search strategies across different problems.