Scaling limits for weakly pinned Gaussian random fields under the presence of two possible candidates (Q904196): Difference between revisions
From MaRDI portal
Set OpenAlex properties. |
Changed an Item |
||
Property / arXiv ID | |||
Property / arXiv ID: 1406.7766 / rank | |||
Normal rank |
Revision as of 18:52, 18 April 2024
scientific article
Language | Label | Description | Also known as |
---|---|---|---|
English | Scaling limits for weakly pinned Gaussian random fields under the presence of two possible candidates |
scientific article |
Statements
Scaling limits for weakly pinned Gaussian random fields under the presence of two possible candidates (English)
0 references
12 January 2016
0 references
Let \({D_N=\{0,1,2,\dots,N\}\times\mathbb{T}_N^{d-1}}\) be the \(d\)-dimensional square lattice and \({D=[0,1]\times\mathbb{T}^{d-1}}\) be its corresponding macroscopic region, where \({\mathbb{T}_N=\mathbb{Z}/N\mathbb{Z}}\) and \({\mathbb{T}=\mathbb{R}/\mathbb{Z}}\). Let \({\varphi=(\varphi_i)_{i\in D_N}}\) be an \(\mathbb{R}\)-valued microscopic field with Dirichlet boundary condition \[ \varphi_i=aN\;\;\text{for}\;\;i\in\{0\}\times\mathbb{T}_N^{d-1},\;\varphi_i=bN\;\;\text{for}\;\;i\in\{N\}\times\mathbb{T}_N^{d-1} \] and \({h^N:D\to\mathbb{R}}\) be its macroscopic counterpart defined as a step function \[ h^N(t)=\frac{1}{N}\varphi_i \;\;\text{for}\;t\in[(i-1/2)/N,(i+1/2)/N)^d, \;i\in D_N. \] For \({\varepsilon>0}\) (strength of the pinning toward 0), consider the Gibbs probability measure on \(\mathbb{R}^{D_N}\) \[ \mu^\varepsilon_N(d\,\varphi)=\frac{1}{Z^\varepsilon_N}e^{-H_N(\varphi)}\prod_{i\in D_N}|\varepsilon\delta_0(d\,\varphi_i)+d\,\varphi_i|, \] where \({Z_N^\varepsilon}\) is the normalizing constant and \({H_N}\) is the Hamiltonian defined as \[ H_N(\varphi)=\frac{1}{2}\sum_{|i-j|=1}(\varphi_i-\varphi_j)^2. \] The topic of the article is the investigation of the asymptotic behavior of \({h^N}\) distributed under \({\mu_N^\varepsilon}\) as \({N\to\infty}\). The main result states that, if \({d\geq3}\), \({\varepsilon>0}\) is sufficiently large and a certain large deviation rate functional \({\Sigma}\) has the same values at both its well defined minimizers \({\hat{h}\neq\bar{h}}\), i.e., \({\Sigma(\hat{h})=\Sigma(\bar{h})}\), then \[ \lim_{N\to\infty}\mu^\varepsilon_N(\|h^N-\hat{h}\|_{L_1(D)}\leq\delta)=1 \] for any \({\delta>0}\). This result extends the one-dimensional result proved in [\textit{E. Bolthausen} et al., Probab. Theory Relat. Fields 143, No. 3--4, 441--480 (2009; Zbl 1180.60077)] and [\textit{T. Funaki} and \textit{T. Otobe}, J. Math. Soc. Japan 62, No. 3, 1005--1041 (2010; Zbl 1197.60090)]. The authors also conjecture that the conditions on \(d\) and \(\varepsilon\) are not necessary for the result.
0 references
Gaussian random fields
0 references
interface model
0 references
pinning
0 references
scaling limits
0 references
large deviations
0 references
minimizers
0 references