arXiv subjects
Publications and source records attributed to Emilio De Santis.
In this article we create a new algorithm for the perfect simulation of the infinite Potts model at a sufficiently small or at a sufficiently high temperature, in particular under the transition phase temperature. We study the model for free boundary conditions and we give some consequences for the constant boundary conditions.
In this paper we consider the problem of determining the law of binary stochastic processes from transition kernels depending on the whole past. These kernels are linear in the past values of the process. They are allowed to assume values close to both 0 and 1, preventing the application of usual results on uniqueness. More precisely we give sufficient conditions for uniqueness and non-uniqueness. In the former case a perfect simulation algorithm is also given.
This paper deals with the problem of perfect sampling from a Gibbs measure with infinite range interactions. We present some sufficient conditions for the extinction of processes which are like supermartingales when large values are taken. This result has deep consequences on perfect simulation, showing that local modifications on the interactions of a model do not affect simulability. We also pose the question to optimize over a class of sequences of sets that influence the sufficient condition for the perfect simulation of the Gibbs measure. We completely solve this question both for the long range Ising models and for the spin models with finite range interactions.
In this paper a general approach for the perfect simulation of a stationary process with at most countable state space is outlined. The process is specified through a kernel, prescribing the probability of each state conditional to the whole past history. We follow the seminal paper of Comets, Fernandez and Ferrari, where sufficient conditions for the construction of a certain perfect simulation algorithm have been given. We generalize this approach by defining backward coalescence times for these kind of processes; this allows us to construct perfect simulation algorithms under weaker conditions.
We introduce and study a class of infinite-horizon non-zero-sum non-cooperative stochastic games with infinitely many interacting agents using ideas of statistical mechanics. First we show, in the general case of asymmetric interactions, the existence of a strategy that allows any player to eliminate losses after a finite random time. In the special case of symmetric interactions, we also prove that, as time goes to infinity, the game converges to a Nash equilibrium. Moreover, assuming that all agents adopt the same strategy, using arguments related to those leading to perfect simulation algorithms, spatial mixing and ergodicity are proved. In turn, ergodicity allows us to prove ``fixation'', i.e. that players will adopt a constant strategy after a finite time. The resulting dynamics is related to zero-temperature Glauber dynamics on random graphs of possibly infinite volume.
We prove, using the random-cluster model, a strict inequality between site percolation and magnetization in the region of phase transition for the d-dimensional Ising model, thus improving a result of [CNPR76]. We extend this result also at the case of two plane lattices Z^2x{0,1} (slabs) and give a characterization of phase transition in this case. The general case of N slabs, with N an arbitrary positive integer, is partially solved and it is used to show that this characterization holds in the case of three slabs with periodic boundary conditions. However in this case we do not obtain useful inequalities between magnetization and percolation probability.
In this paper we consider two-point unilateral Markov fields on a two-dimensional lattice as considered by Pickard, Galbraith and Walley. We show that, under various ergodicity conditions, they can be perfectly simulated in the stationary state on any finite window. The techniques which are used connect perfect simulation with oriented percolation through suitable coupling constructions.
We analyze several aspects of a class of simple counting processes, that can emerge in some fields of applications where the presence of a change-point occurs. Under simple conditions we, in particular, prove a significant inequality for the stochastic intensity.
We study the Swendsen-Wang dynamics for disordered non ferromagnetic Ising models on cubic subsets of the hypercubic lattice Z^d and we show that for all small values of the temperature parameter T the dynamics has a slow relaxation to equilibrium (it is torpid). Looking into this dynamics from the point of view of the Markov chains theory we can prove that the spectral radius goes to one when the size of the system goes to infinity. This means that, if we want to use the Swendsen-Wang dynamics for a computer simulation, we have a slow convergence to the stationary measure in low temperature. Also it is a good example of a non-local dynamics that relaxes slowly to the equilibrium measure.
We consider stochastic processes, S^t \equiv (S_x^t : x \in Z^d), with each S_x^t taking values in some fixed finite set, in which spin flips (i.e., changes of S_x^t) do not raise the energy. We extend earlier results of Nanda-Newman-Stein that each site x has almost surely only finitely many flips that strictly lower the energy and thus that in models without zero-energy flips there is convergence to an absorbing state. In particular, the assumption of finite mean energy density can be eliminated by constructing a percolation-theoretic Lyapunov function density as a substitute for the mean energy density. Our results apply to random energy functions with a translation-invariant distribution and to quite general (not necessarily Markovian) dynamics.