Entropy dissipation estimates in a zero-range dynamics (Q2641900)

From MaRDI portal





scientific article
Language Label Description Also known as
default for all languages
No label defined
    English
    Entropy dissipation estimates in a zero-range dynamics
    scientific article

      Statements

      Entropy dissipation estimates in a zero-range dynamics (English)
      0 references
      0 references
      0 references
      17 August 2007
      0 references
      The complete graph zero-range dynamics is considered, i.e., the authors study the following continuous time Markov chain. The state space \(\Omega_L = \mathbb{Z}_+^{L}\) where \(L\) is a positive integer and a configuration \(\eta \in \Omega_L\) is interpreted as an occupation number vector, that is \(\eta_x\) is a number of particles at \(x\in V_L=\{1,\ldots,L\}\) (\(L\geq 2\)). For each \(x\in V_L\) there exists the rate function \(c_x:\mathbb{Z}_+ \to \mathbb{R}\) such that \(c_x(0)=0\) and \(c_x(n) >0\) for any \(n \in \mathbb{N}\). Every vertex \(x\in V_L\) waits an exponentially distributed time with mean \(1/c_x\) before one particle is moved from \(x\) to a uniformly chosen vertex of \(V_L\), one can specify the construction by means of the explicit formula for the generator \(\mathcal{L}\). Let \(\mu_L\) be the product measure \(\prod_{x\in V_L} \mu_x\) where \(\mu_x(0)= 1/Z_x\) and \(\mu_x(n) = \frac{1}{Z_x}\prod_{k=1}^n \frac{1}{c_x(k)}\). Introduce \(\nu = \nu_{L,N}\) as the probability \(\mu_L\) conditioned on the event \(N= \sum_{x\in V_L}\eta_x\). For a function \(f\geq 0\) set \(\text{Ent}_{\nu} = \nu[f\log f] - \nu[f] \log \nu[f]\) where \(\nu[f]:= \int f d\nu\). Put \(f_t := e^{t\mathcal{L}}f\), \(t\geq 0\). The goal is to guarantee the existence of a (the best possible) constant \(\gamma = \gamma(L,N)\) such that \(\text{Ent}_{\nu}(f_t) \leq E^{-t/\gamma} \text{Ent}_{\nu}(f)\). Under certain conditions imposed on the rate function (in particular, one uses the homogeneous rates such that for some \(c:\mathbb{Z}_+ \to \mathbb{R}_+\) one has \(c_x(n)=c_y(n)=c(n)\) for all \(x,y \in V_L\) and \(n \in \mathbb{Z}_+\)) it is proved that \(\sup_{L\geq 2} \sup_{N\geq 1} \gamma(L,N) < \infty\). Thus for the standard model with rates increasing at infinity the entropy dissipation estimates, uniform over the number of particles and the number of vertices, are established. The conjectures and suggestions concerning the study of inhomogeneous rates are provided.
      0 references
      zero-range dynamics
      0 references
      entropy dissipation
      0 references
      modified logarithmic Sobolev inequalities
      0 references
      0 references
      0 references
      0 references
      0 references

      Identifiers