Pinning for the critical and supercritical membrane model

Abstract

The membrane model is a Gaussian interface model with a Hamiltonian involving second derivatives of the interface height. We consider the model in dimension $\mathsf{d}\ge 4$ under the influence of $\delta$-pinning of strength $\epsilon$. It is known that this pinning potential manages to localize the interface for any $\epsilon >0$. We refine this result by establishing the $\epsilon$-dependence of the variance and of the exponential decay rate of the covariances for small $\epsilon$ (similar to the corresponding results for the discrete Gaussian free field by Bolthausen-Velenik). We also show the existence of a thermodynamic limit of the field. These conclusions improve upon earlier works by Bolthausen-Cipriani-Kurt and by Sakagawa. The problem has similarities to the homogenization of elliptic operators in randomly perforated domains, and our proof takes inspiration from this connection. The main new ideas are a correlation inequality for the set of pinned points, and a probabilistic Widman hole filler argument which relies on a discrete multipolar Hardy-Rellich inequality and on a multi-scale argument to construct suitable test functions.

Corrected some typos

References in corpus (8)

• Localization and delocalization of random interfaces
• Royen's proof of the Gaussian correlation inequality
• Pinning and wetting transition for (1+1)-dimensional fields with Laplacian interaction
• The Method of Reflections, Homogenization and Screening for Poisson and Stokes Equations in Perforated Domains
• Maximum and entropic repulsion for a Gaussian membrane model in the critical dimension
• Scaling limits of $(1+1)$-dimensional pinning models with Laplacian interaction
• Extremes of some Gaussian random interfaces
• Exponential decay of covariances for the supercritical membrane model

Cited by in corpus (1)

• Thermodynamic and Scaling Limits of the non-Gaussian Membrane Model