We study a system of N layers with a Kac horizontal interaction of parameter γ > 0 and a Kac vertical interaction of parameter γ 1/2 . We shall prove that the limit free energy functional is the rate function of the large deviations of the Gibbs measure (of a canonical constrained magnetization). The limit free energy functional is achieved as a -limit for γ → 0 for magnetizations with fixed average. Among all such magnetizations there exists a quasiconstant magnetization that minimizes the energy.Communicated by Raffaele Esposito. MSC2010: 82B24. 267 268 MICHELE ALEANDRI AND VENANZIO DI GIULIO between nearest neighbor horizontal lines (say (x, i) and (y, i + 1)) interact via the Kac potential λJ γ 1/2 (x, y), λ > 0; that is, the vertical interaction is much more local than the horizontal one.We study the mesoscopic limit γ → 0. The mesoscopic state of the system is a collection m ≡ {m(r, i) : r ∈ [0, ], i = 1, . . . , N } of measurable functions with values in [−1, 1]. Its statistical properties are then described by a free energy functional F(m). According to the Gibbs theory such a functional is the limit as γ → 0 of −1/β times the log of a constrained partition function where the spin configurations are required to be "close" to the mesoscopic state m (this involves a coarse grain procedure which is specified in Section 2). This is not as in the classical Lebowitz-Penrose [1966;Penrose and Lebowitz 1971] procedure because there are two scales, γ −1 for the horizontal interaction and γ −1/2 for the vertical one. Thus, there could be oscillations on the scale γ −1/2 which do not appear in m because the latter is defined by averages over ≈ γ −1 but which could affect the free energy of m. These oscillations actually do not occur if λ is small; indeed by Theorem 4.1 the optimal profile is quasiconstant on the scale γ −α with α ∈ (0, 1). However, if λ is large enough we can provide an example where such a phenomena occurs.The paper is organized as follows. In Section 2 we introduce the microscopic and mesoscopic models and enunciate the main results. In Section 3 we introduce the coarse graining procedure used to prove the Lebowitz-Penrose limit. In Section 4 we prove a key result, that is, Theorem 4.1, in which we provide a technique to minimize the free energy. This theorem is needed to prove the main results in Section 2. In Section 5 we prove the Lebowitz-Penrose limit for our model. In Section 6 we prove the -limit result. The proofs of Theorem 2.4 and Proposition 3.1 are deferred to Appendix A. In Appendix B, finally, we illustrate the case in which Theorem 2.3 fails for the parameter λ large enough.Similar model have been studied in [Cassandro et al. 2016;Fontes et al. 2014;. A numerical investigation of the mesoscopic limit for lattice gas model was also recently tackled in [Colangeli et al. 2016;.This work is the first step of a research program pointed towards the characterization of the surface tension associated to free energy in the thermodynamic limit.
Model and main resultsWe consider an Ising spin ...