--------------------------------------------------------------------
   COLLOQUIUM OF THE LABORATORY FOR COMPUTER DESIGN OF MATERIALS
                   School of Computational Sciences
                           (CSI 898-Sec 001)
--------------------------------------------------------------------

         Statistical Mechanics Partition Functions with 
             Modified Potential Energy Landscapes

                        Mark Fleischer
   Johns Hopkins University Applied Physics Laboratory, Laurel, MD

A new approach for overcoming the broken ergodicity problem in Markov 
Chain Monte Carlo (MCMC) simulations of complex systems will be 
presented in this talk.

Unlike other approaches that modify the Metropolis MCMC algorithm, our
approach leaves the MCMC features essentially unchanged and instead 
modifies the potential energy landscape of the system through the use 
of transformation functions.  These transformation functions have 
their roots in certain scale invariance properties of non-extensive
statistical mechanics.  By using a generalized form of non-extensive
entropy, a family of transformation functions provide useful 
information that permits extrapolations from the modified potential energy 
landscape to the original, unmodified potential landscape.  With this innovative
approach we obtain estimates of the exact Boltzmann Partition Function, 
bounded by small errors bars. Theoretical results and computational 
examples will be provided.
                        
                      Monday , April 18, 2005
                                  4:30 pm
               Room 206, Science & Tech. I, Fairfax Campus

Refreshments will be served at 4:15 PM.
----------------------------------------------------------------------
Find the schedule at http://www.scs.gmu.edu/lcdm/seminar/schedule.html
--------------------------------------------------------------------