Generalized Ensemble Simulations of Complex Systems

Dr. Bernd A. Berg
Department of Physics and
School of Computational Science and Information Technology
Florida State University, Tallahassee, FL 32306, USA

Monte Carlo (MC) simulations are an indispensable tool for  the investigation of physical models. The most efficient MC weights for the calculation of physical, canonical expectation  values are not necessarily those of the canonical ensemble, but  the use of suitably generalized ensembles can lead to much faster convergence. Although not realized by nature, these ensembles can be implemented on computers.

 In recent years the generalized ensemble approach has in particular  been studied for the simulation of complex systems. For these  systems it is typical that conflicting constraints lead to free  energy barriers, which fragment the configuration space. Examples of major interest are spin glasses and proteins. In my overview I will first comment on the strengths and weaknesses of a few major approaches, including parallel tempering, multicanonical and transition variable methods. Subsequently, selected examples from applications to spin glasses and proteins will be presented.
 
 For a review of the generalized ensemble approach in the context of protein folding see Hansmann and Okamoto [1]. The multicanonical approach, to some extent along with parallel  tempering and transition variable methods, is reviewed in  Ref. [2]. The featured applications will include Helix-coil  transitions of amino-acid homo-oligomers [3] and a new analysis  of the Parisi overlap distribution from a simulation of the 3d Edwards-Anderson Ising spin glass [4]

References

Zeit: Montag, 05. Nov. 01 um 14.15 Uhr, Tee/Kaffee ab 14.00 Uhr
Ort:  ZIB, Takustr. 7, 14195 Berlin-Dahlem
Raum:  Seminarraum 2006 (EG, im Rundbau)