Swarm Intelligence Steers a Global Minima Search of Clusters Bound on Carbon Nanostructures

Abstract

A detailed exploration of the energy landscape is a necessary criterion for an efficient structure prediction. Although computational power has increased many-folds in recent decades, a full analysis of the potential energy surfaces (PESs) for complex chemical systems is still computationally elusive. Locating the most stable geometry of a system from a highly complex PES necessitates the use of metaheuristic techniques. Particle swarm optimization (PSO), a nature-inspired metaheuristic technique falling under the general ambit of artificial intelligence, has proven to be a promising approach for structure prediction. Through social networking, the swarm attains a collective intelligence leading to an effective search of the PESs for the most stable configurations. The PESs associated with the noncovalent interactions involving carbon nanostructures and atomic or molecular clusters are often rugged, with a large number of local minima hindering the tracking down of the global minima. In this Perspective, we demonstrate that an amalgamation of PSO with the continuum approximation, an approximation that exploits the symmetry of carbon nanostructures, serves as a robust methodology for unraveling the interactions of Lennard-Jones clusters with carbon nanostructures. The putative global minima geometries obtained from our methodology can provide fruitful insights on the structural and energetic aspects of cluster binding on carbon nanostructures and can offer excellent initial configurations for first-principles calculations.

High-Pressure Synthesis of Bulk Cobalt Cementite, Co3C

“…An exhaustive survey of the high-pressure phase spacethat is, the determination of the energy of every possible structurewould rely on the calculation of an intractably large potential energy surface (PES). Since this is currently not possible, various groups have devised much more efficient methods for searching for global minima on the PES. − We used the AIRSS method , coupled with the CASTEP first-principles calculations package to survey the cobalt–carbon system under high pressures …”

Accelerating the prediction of large carbon clusters via structure search: Evaluation of machine-learning and classical potentials

“…Thus, to target cluster sizes in the C200 to C720 range, we employ box sizes of 100 Å and an initial sphere radius of 9.5 Å. As known from previous studies 49,50 , in this range size the fullerene/cage-like structures with high symmetry (T h /T d or I h ) are expected to form. Therefore we covered this possibility by enforcing all 24 symmetry operators while generating the input structures.…”

Performance of Particle Swarm Optimization in Predicting the Orientation of π-Conjugated Molecules Inside Carbon Nanotubes Compared with Density Functional Theory Calculations

“…The positions of particles are updated until a specified number of iterations is completed. − Call et al first applied PSO to obtain the global minima of the Lennard-Jones (LJ) 26-atom cluster, Si 2 H 5 – , and triply hydrated OH – . After the first application, various systems, such as metal clusters, crystal structures, , N 2 clusters on an sp 2 graphene flake, and CNTs containing a carbon ring or LJ clusters , were studied using PSO techniques.…”

“…Although several PSO studies have been conducted, − to the best of our knowledge, no literature has discussed the accuracy of PSO in predicting nanometer-sized structures compared with those obtained through DFT calculations. In this direction, we applied the PSO method to investigate the orientation of triiodobenzene (BzI 3 ) inside an armchair ( m , m ) CNT (BzI 3 @( m , m )).…”