Basin-hopping and beyond: Global optimization and energy landscape exploration in molecular systems
Authors: Jacob Stevenson and Victor Rühle
Global optimization is an important tool for structure prediction in computational physics, chemistry, and biology. We will describe the global optimization method basin-hopping (
scipy.optimize.basinhopping), and how we use it to predict the structure of atomic and molecular clusters, biomolecular systems and crystals.
Furthermore, we present a computational framework which combines basin-hopping with other methods to explore the potential and free energy landscape of molecular systems. We use this framework to build a database of stable, low energy structures and to obtain transition states and minimum-energy paths connecting them. From the database of minima and their connectivity, we compute reaction pathways and thermodynamic properties.
All methods are available in a python based framework under an open source software license. For our research, we speak in terms of minimizing the energy of a molecular structure, but our framework is applicable to anyone interested in exploring the landscape of a smooth scalar function.