Density of states and numerical simulations : from Statistical Mechanics to Gauge Theories

Orateur : Biagio LUCINI (Swansea University)

Determining the density of states through a Monte Carlo method allow us to rephrase numerical simulations into a framework that circumvents importance sampling and with it some of the drawbacks of importance sampling methods. In this work, we shall present a recently proposed algorithm based on a non-Markovian process which has been conceived to determine efficiently continuous density of states. Using as prototype the Potts model and the U(1) Lattice Gauge Theory system, we shall show that the method is very effective at avoiding ergodicity problems related to strong metastabilities at first order phase transition points. We then present more preliminary results on the direct computation of partition functions, on the avoidance of topological trapping in systems where topological objects play a crucial role in the dynamics and on the “solution” of the sign problem.

Télécharger dans mon agenda