Improving Atomistic Crystal Structure Prediction Searches through Symmetric Initialization

Patrick Avery
SUNY Buffalo

Unique symmetry distributions have been observed in different classes of extended systems. Less than 1% of inorganic crystals belong to the P1 space group (or in other words, they contain no symmetry). This demonstrates that symmetry is very important in extended matter. In computational materials science, new crystals are often discovered through crystal structure prediction (CSP) methods, which often involve an initial step of randomly generating crystals. Constraining these randomly generated crystals to be of specific space groups has been shown to often improve the rate of finding the global minimum structure by a significant amount. On the other hand, in experiment, some synthesized crystal structures are difficult to elucidate due to the conditions of the experiment or the presence of lightweight atoms; however, sometimes information such as their composition and space group are known. In this presentation, I will discuss different methods that have been used to perform symmetric initialization - including our group's program that was developed for this purpose: RandSpg.

Presentation (PowerPoint File)

Back to Workshop I: Optimization and Optimal Control for Complex Energy and Property Landscapes