Diffusive long-time behavior of Kawasaki dynamics (Q1767554)

From MaRDI portal
Revision as of 01:00, 27 July 2023 by Importer (talk | contribs) (‎Created a new Item)
(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
scientific article
Language Label Description Also known as
English
Diffusive long-time behavior of Kawasaki dynamics
scientific article

    Statements

    Diffusive long-time behavior of Kawasaki dynamics (English)
    0 references
    0 references
    0 references
    0 references
    8 March 2005
    0 references
    The authors consider a semigroup associated with a Kawasaki dynamics in \(\mathbb Z^d\). The model is a lattice gas in \(\mathbb Z^d\) whose phase space is given by \(\Omega=S^{\mathbb Z^d}\), where \(S=\{0,1\}\). On the phase space a Hamiltonian \(H(\eta)\) is defined generated by a short range, translation invariant bounded interaction \(J(\eta)\). Under a suitable assumption on \(J\), called by the authors condition (USM), which is implied, except for a uniformity condition, by \textit{R. L. Dobrushin} and \textit{S. B. Shlosman}'s complete analyticity [J. Stat. Phys. 46, No. 5/6, 983--1014 (1987; Zbl 0683.60080)], it is possible to prove the existence of only one Gibbs state \(\mu\) associated with the interaction. The dynamics on \(\Omega\) is given by a formal generator \(\mathcal L\) defined in the folloving way: \[ (\mathcal L F)(\sigma)=\sum_{<x,y>}c_{x,y}(\sigma)\nabla_{x,y}f, \] where the sum goes over the nearest neighbors and \(\nabla_{\cdot,\cdot}\) is the discreted gradient. The assumptions on the transition rates \(c_{\cdot,\cdot}(\cdot)\) of the process are: a) finite range, b) detailed balance, c) positivity and boundedness. Under all these assumptions the authors prove the \(L_2\)-convergence of the evolution. More precisely, let \(P_t\) be the semigroup generated by \(\mathcal L\), then for any \(\varepsilon\) and for all local functions \(f\) on \(\Omega\) there is a positive constant \(A(\varepsilon,f)\) such that \[ \mu \left [ (P_tf-\mu(f))^2\right]\leq A(\varepsilon,f)t^{-(d/2 -\varepsilon)}. \] This result is proved for \(d\geq 3\), because the cases \(d=1,2\) were studied by \textit{N. Cancrini} and \textit{F. Martinelli} [J. Math. Phys. 41, No. 3, 1391--1423 (2000; Zbl 0977.82031)].
    0 references
    convergence to the invariant measure
    0 references

    Identifiers