Genetic algorithm optimization of atomic clusters

Conference · Tue Dec 31 04:00:00 UTC 1996

OSTI ID:505392

Morris, J R; Deaven, D M; Ho, K M; Wang, C Z; Pan, B C; Wacker, J G; Turner, D E ^[1]

---

1. Ames Lab., IA (United States)

The authors have been using genetic algorithms to study the structures of atomic clusters and related problems. This is a problem where local minima are easy to locate, but barriers between the many minima are large, and the number of minima prohibit a systematic search. They use a novel mating algorithm that preserves some of the geometrical relationship between atoms, in order to ensure that the resultant structures are likely to inherit the best features of the parent clusters. Using this approach, they have been able to find lower energy structures than had been previously obtained. Most recently, they have been able to turn around the building block idea, using optimized structures from the GA to learn about systematic structural trends. They believe that an effective GA can help provide such heuristic information, and (conversely) that such information can be introduced back into the algorithm to assist in the search process.

• Research Organization: Ames Lab., IA (United States)
• Sponsoring Organization: USDOE Office of Energy Research, Washington, DC (United States)
• DOE Contract Number: W-7405-ENG-82
• OSTI ID: 505392
• Report Number(s): IS-M--868; CONF-9610311--1; ON: DE97006804
• Country of Publication: United States
• Language: English

Related Subjects

66 PHYSICS
ALGORITHMS
ATOMIC CLUSTERS
ENERGY LEVELS
LENNARD-JONES POTENTIAL
MOLECULAR STRUCTURE
OPTIMIZATION
PERFORMANCE