Abstract
The large potential energy barriers separating local minima on the potential energy surface of cluster systems pose serious problems for optimization and simulation methods. This article discusses algorithms for dealing with these problems. Lennard-Jones clusters are used to illustrate the important issues. In addition, the complexities in going from one-component to binary Lennard-Jones clusters are explored.
| Original language | English |
|---|---|
| Pages (from-to) | 1-23 |
| Number of pages | 23 |
| Journal | Computer Physics Communications |
| Volume | 145 |
| Issue number | 1 |
| DOIs | |
| State | Published - May 1 2002 |
| Externally published | Yes |
Funding
This research was supported with a grant from the National Science Foundation and with support through a NPACI partnership at NCSA. The calculations were carried out on the NT supercluster at NCSA, the IBM SP (RS/6000) computer at SDSC, a Pentium cluster at the Pittsburgh Supercomputing Center, and the IBM RS6000 workstation cluster in the University of Pittsburgh’s Center for Molecular and Materials Simulations (CMMS). The computers in CMMS were funded by grants from the National Science Foundation and IBM. We thank David Wales for many helpful discussions and for access to his GMIN basin-hopping global minimization program, and Dave Freeman for valuable discussions about the parallel tempering algorithm. We also thank Mark Miller for access to his programs for constructing disconnectivity tree diagrams.
| Funders | Funder number |
|---|---|
| National Science Foundation | |
| International Business Machines Corporation |
Keywords
- Clusters
- Global optimization
- Monte Carlo simulations
- Parallel computing