Generating derivative structures at a fixed concentration

We present an algorithm for generating derivative superstructures for large unit cells at a fixed concentration. The algorithm is useful when partial crystallographic information of an ordered phase is known. This work builds on the previous work of Hart and Forcade [Phys. Rev. B 77 224115, 2008; Phys. Rev. B 80 014120, 2009]. This extension of the original algorithm provides a mapping from atomic configurations to consecutive integers when only a subset (fixed concentration) of all possible configurations is under consideration. As in the earlier algorithm, this mapping results in a minimal hash table and perfect hash function that enables an efficient method for enumerating the configurations of large unit cells; the run time scales linearly with the number of symmetrically-distinct configurations. We demonstrate the algorithm for a proposed structure in the Ag–Pt system comprising 32 atoms, with stoichiometry 15:17, a configuration space of ∼400,000.

► An order-N algorithm for generating large derivative structures is developed.
► The algorithm is useful for bulk and surface alloys, oxides, hydrogen storage, etc.
► The algorithm extends previous work to much larger cases.