A Review of Global Optimization Methods for Molecular Structures: Algorithms, Applications and Perspectives.

This review presents a comprehensive overview of global optimization techniques applied to the prediction of chemical structures, including molecular conformations, crystal polymorphs, and reaction pathways. These approaches typically involve a two-step process: a global search to identify candidate structures, followed by local refinement to determine the most stable configurations. Global optimization methods are commonly grouped into two categories, known as stochastic and deterministic methods, based on their exploration strategies and underlying theoretical principles. A historical perspective highlights the progression of these methods, from early foundational algorithms to more advanced and efficient modern techniques. For each category, key algorithmic frameworks are outlined, widely used software tools are discussed, and representative applications are examined, such as conformer sampling, cluster structure prediction, and surface adsorption. The review concludes by considering future directions, including the integration of accurate quantum methods, the development of flexible hybrid algorithms, and the use of quantum computing to address increasingly complex chemical problems.