On the Ising model with random boundary condition (Q2487795)

The authors consider the problem of chaotic size dependence for the 2d-Ising model with random boundary conditions, sampled from an 2nd symmetric random field. The work is motivated by the Newman--Stein metastate approach to finite--range spin glasses. In a previous work by \textit{A. C. D. van Enter, I. Medved'} and \textit{K. Netocny} [Markov Process. Relat. Fields 8, No. 3, 479--508 (2002; Zbl 1010.82016)], the model was considered in the particular case where the coupling with the boundary conditions is sufficiently weak, while in the present paper it is allowed to be of the same order of magnitude as the bulk couplings. The main result of the present paper is that, if the temperature is low enough and if the thermodynamic limit is approached along a sufficiently sparse sequence of cubes of sides \(k_N\), then on a set of disorder realizations of measure~1 the finite-volume Gibbs measure \(\mu_{\Lambda(k_N)}\) oscillates between the Ising pure measures \(\mu^+\) and \(\mu^-\). In particular, this means that the limit of the Gibbs measure does not exist (almost surely) and that mixed phases are not observed (along these sub-sequences). In order to exclude mixtures as possible limit points, a detailed multi-scale contour analysis is performed. Moreover, a number of conjectures are put forward by the authors: in dimensions \(d\geq4\), they expect that the set of (weak) limit points of finite-volume Gibbs measures \( \{\mu_{\Lambda(N)}\}_{N=1,2,\ldots}\) coincides with \(\{\mu^-,\mu^+\}\). Note that this would be a much stronger result, since in this conjecture one considers the whole sequence of finite-volume Gibbs measure, and not just a sparse sub-sequence of it. On the other hand, in dimensions 2 and 3 they expect that the set of limit points of \( \{\mu_{\Lambda(N)}\}_{N=1,2,\ldots}\) should include all translation-invariant infinite volume Gibbs measures, including mixtures (but not interface phases).