[Paper Review] Random skew plane partitions and the Pearcey process

Andreĭ Okounkov, Nicolai Reshetikhin|ArXiv.org|Mar 23, 2005

Random Matrices and Applications7 references4 citations

TL;DR

This paper studies random skew plane partitions weighted by $q^{\text{vol}}$ in the $q \to 1$ limit, using Schur process techniques to derive exact contour integral formulas for correlation functions. It establishes universal asymptotic behavior: sine-kernel in the bulk, Airy kernel along the frozen boundary, and Pearcey kernel near cusps—providing the first rigorous derivation of the Pearcey process in this context.

ABSTRACT

We study random skew 3D partitions weighted by $q^{ extup{vol}}$ and, specifically, the $q o 1$ asymptotics of local correlations near various points of the limit shape. We obtain sine-kernel asymptotics for correlations in the bulk of the disordered region, Airy kernel asymptotics near a general point of the frozen boundary, and a Pearcey kernel asymptotics near a cusp of the frozen boundary.

Motivation & Objective

• To understand the $q \to 1$ asymptotics of local correlation functions in random skew 3D partitions weighted by $q^{\text{vol}}$.
• To identify universal microscopic correlation structures—sine, Airy, and Pearcey kernels—based on the macroscopic geometry of the limit shape.
• To establish that the Pearcey kernel governs correlations near cusps of the frozen boundary, confirming a conjecture in random matrix theory.
• To demonstrate that the frozen boundary is an algebraic curve with exactly one cusp per exterior corner of the inner shape $\mu$.
• To extend the universality hypothesis for local statistics in random surface models to the cusp singularity case.

Proposed method

• Model the random skew 3D partition as a Schur process, enabling exact contour integral representations of correlation functions.
• Use saddle-point analysis on the contour integral formulas to extract asymptotics in the $q \to 1$ (thermodynamic) limit.
• Identify the nature of critical points in the saddle-point analysis: simple, double, or triple, corresponding to sine, Airy, and Pearcey kernels respectively.
• Relate the geometry of the limit shape to the singularity structure of the frozen boundary via the critical point analysis.
• Apply Wick’s lemma to express correlation functions as determinants of a kernel matrix derived from the Schur process.
• Utilize symmetry transformations of the integral kernel to derive reflection symmetries in the correlation functions, confirming consistency with tiling invariance.

Experimental results

Research questions

• RQ1 What is the universal microscopic correlation structure near a cusp of the frozen boundary in random skew plane partitions?
• RQ2 How do the local correlations behave in the bulk of the disordered region as $q \to 1$?
• RQ3 What is the precise asymptotic form of the correlation kernel near a general point of the frozen boundary?
• RQ4 How is the singularity type (cusp, turning point) of the frozen boundary related to the critical point structure of the saddle-point analysis?
• RQ5 Can the Pearcey kernel emerge as the universal scaling limit near a cusp in a discrete random tiling model?

Key findings

• Near a cusp of the frozen boundary, the local correlation functions, when suitably scaled, converge to the extended Pearcey kernel, confirming the universality of this kernel in this context.
• In the bulk of the disordered region, the correlation functions exhibit sine-kernel asymptotics, consistent with the incomplete beta kernel and discrete sine kernel limits.
• Along the frozen boundary at a general point, the correlation functions scale to the extended Airy kernel, as predicted by universality hypotheses.
• The frozen boundary is an algebraic curve with exactly one cusp per exterior corner of the inner shape $\mu$, and its singularities are fully classified by the saddle-point critical point types.
• The asymptotic analysis reveals that the Pearcey kernel arises when the saddle point is a triple critical point, linking the singularity type to the kernel universality class.
• The symmetry of the correlation kernel under $t \to -t$ and variable reflection confirms a microscopic invariance of the tiling model under time reversal in the $t$-direction.