A Fukui function-guided genetic algorithm. Assessment on structural prediction of Sin (n = 12–20) clusters

Osvaldo Yañez, Alejandro Vásquez-Espinal, Diego Inostroza, Lina Ruiz, Ricardo Pino-Rios, William Tiznado

Research output: Contribution to journalArticlepeer-review

5 Citations (Scopus)

Abstract

Theoretical studies are essential for the structural characterization of clusters, when it comes to rationalize their unique size-dependent properties and composition. However, the rapid growth of local minima on the potential energy surface (PES), with respect to cluster size, makes the candidate identification a challenging undertaking. In this article, we introduce a hybrid strategy to explore the PES of clusters. This proposal involves the use of a biased initial population of a genetic algorithm procedure. Each individual in this population is built by assembling small fragments, according to the best matching of the Fukui function. The performance of a genetic algorithm procedure. The performance of the method is assessed on the PES exploration of medium-sized Sin clusters (n = 12–20). The most relevant results are: (a) the method converges at almost half of the time used by the canonical version of the GA and, (b) in all the studied cases, with the exception of Si13 and Si16, the method allowed to identify the global minimum (GM) and other important low-lying structures. Additionally, the apparent deficiency of the proposal to identify the GM was corrected when a Si atom, or other low-lying isomers, were considered to build the clusters.

Original languageEnglish
Pages (from-to)1668-1677
Number of pages10
JournalJournal of Computational Chemistry
Volume38
Issue number19
DOIs
Publication statusPublished - 15 Jul 2017

Keywords

  • Fukui function
  • clusters
  • genetic algorithm
  • potential energy surface exploration

ASJC Scopus subject areas

  • Chemistry(all)
  • Computational Mathematics

Fingerprint

Dive into the research topics of 'A Fukui function-guided genetic algorithm. Assessment on structural prediction of Si<sub>n</sub> (n = 12–20) clusters'. Together they form a unique fingerprint.

Cite this