Edge statistics for Lozenge tilings of polygons. I: Concentration of height function on strip domains (Q6145692)

A central feature of random lozenge tilings is that they exhibit boundary-induced phase transitions. Depending on the shape of the domain, they can admit frozen regions, where the associated height function is flat almost deterministically, and liquid regions, where the height function appears more rough and random, see [\textit{V. Gorin}, Lectures on random lozenge tilings. Cambridge: Cambridge University Press (2021; Zbl 1479.52001)]. The interface or edge is the boundary between frozen and liquid regions, and in the limit it converges to a non-random curve, the arctic curve. Before the limit, it was predicted that the fluctuations of tiling boundary curves are of order \(n^{1/3}\) and \(n^{2/3}\) in the directions transverse and parallel to their limiting trajectories, respectively. After taking appropriate scaling limit, they converge to the \(Airy_2\) process, a universal scaling limit (see [\textit{K. Johansson}, in: Current developments in mathematics 2016. Papers based on selected lectures given at the current development mathematics conference, Harvard University, Cambridge, MA, USA, November 2016. Somerville, MA: International Press. 47--110 (2018; Zbl 1423.60167)]). This prediction has been established for various tiling models on specific domains. This paper studies uniformly random lozenge tilings of strip domains. This is equivalent to a family of \(m\) non-intersecting random Bernoulli walks conditioned to start and end at specified locations. Under the assumption that the limiting arctic boundary has at most one cusp and some other technical assumptions, the author proves a concentration estimate for the tiling height functions and arctic boundaries on such domains. An informal formulation of this result is provided as follows: Theorem 1.1. For uniformly random lozenge tilings of strip domains (equivalently, non-intersecting random Bernoulli bridges), with limiting arctic boundary containing at most one cusp, it holds that for any \(\delta>0\) the tiling height function is within \(n^{\delta}\) of its limit shape, and the tiling arctic boundary is within \(n^{1/3+\delta}\) of its limit shape with overwhelming probability. The above concentration result is optimal up to the \(n^{\delta}\) factor, and it will be used in Part II of this series [\textit{A. Aggarwal} and \textit{J. Huang}, ``Edge statistics for lozenge tilings of polygons. II: Airy line ensemble'', Preprint, \url{arXiv:2108.12874}] to prove the edge universality for random lozenge tilings. The edge statistics of simply-connected polygonal domains, subject to a technical assumption on their limit shape, converge to the \textit{Airy} line ensemble at any point around the arctic boundary. The concentration results imply that particles of non-intersecting random Bernoulli walks conditioned to start and end at specified locations strongly concentrate around their classical locations. The remainder of this paper is organized as follows. Section 2 defines the model, the associated variational problem, and state the main results. Section 3 collects some properties of complex slopes associated with tilings of double sided trapezoid domains with general boundary height function. Section 4 introduces a weighted version of non-intersecting Bernoulli bridges, and use it to study tilings on the doublesided trapezoid domains. The concentration estimates for weighted non-intersecting Bernoulli bridge model is proved in Section 5 assuming Proposition 5.5 below. Finally, Proposition 5.5 is proved in Sections 6 and 7 by analyzing the dynamical loop equations. In Appendix A, the author checks that thin slices of polygonal domains provide examples of strips with limiting continuum height function satisfies the assumptions. In Appendices B, C and D, the author gives the proofs of some propositions from Section 3.