Scaling limits of equilibrium wetting models in \((1+1)\)-dimension (Q2487018): Difference between revisions

For \(\varepsilon\geq 0\) and \(N\in\mathbb N\) let \[ P_{N,\varepsilon}(dx) := \frac{1}{Z_{N,\varepsilon}}\exp\left(-H_N(x)) \prod_{i=1}^N(dx_i^+ + \varepsilon \delta_0(dx_i)\right), \] where \(dx^+\) is the Lebesgue measure on \(\mathbb R_+\), \(\delta_0\) is the Dirac measure concentrated in 0, \(Z_{N,\varepsilon}^a\) is the normalizing constant and for \(x=(x_1,\ldots, x_N)\in\mathbb R_+^N\), \(H_N(x)\) is the free or the constrained Hamiltonian associated with a potential \(V\) such that \(\exp(-V(\cdot))\) is continuous, \(V(0)<\infty\), \(\int_{\mathbb R} ye^{-V(y)}dy =0\), \(\int_{\mathbb R} y^2 e^{-V(y)}dy <\infty\) and the \(n\)th convolution of \(e^{-V}\) is bounded for some \(n\geq 1\). \textit{Y. Isozaki} and \textit{N. Yoshida} [Stochastic Processes Appl. 96, No. 2, 261--284 (2001; Zbl 1058.60091)] showed that a similar model based on the nearest-neighbor, symmetric random walk, displays a phase transition at a critical value of the reward \(\varepsilon_c>0\). In this paper, similar Brownian scalings are obtained but the techniques are very different. In the subcritical regime, the path converges in distribution to the Brownian meander (resp., the Brownian excursion in the constrained case). In the critical regime, there is convergence to the reflecting Brownian motion (resp. to the reflecting Brownian bridge) and for \(\varepsilon>\varepsilon_c\), there is weak convergence to the constant function taking the value zero. Without rescaling the paths, the following stronger result is proved in the subcritical regime: the set of dry sites (where \(x_i=0\)) is reduced to a finite number of points all at a finite (microscopic) distance from 0 (resp. from \(\{0,N+1\}\) in the constrained case). In the supercritical regime, there is convergence to the law of a finitely recurrent irreducible Markov chain on \(\mathbb R_+\).