From alternating sign matrices to the Gaussian unitary ensemble (Q461427): Difference between revisions

Alternating sign matrices (ASM) are matrices with entries in \(\{0,1,-1\}\) whose row and column sums equal 1 and nonzero entries alternate in sign along rows and columns. They represent 6-vertex models. The purpose of the paper is to derive the properties of uniform random ASMs near their boundaries as their size goes to infinity. The main theorem is a statement about the first \(k\) rows, but by symmetry, `first' can be replaced by `last' and/or `rows' by `columns'. The result states that asymptotically with high probability the number of \(-1\) entries in row \(k\) is \(k-1\) (interlaced with as many \(1\)s). Moreover, the random vector \(\sqrt{{8 \over 2N}}(\eta(N)_j^i-N/2)\) of size \(N\), where \(\eta(N)_j^i\) is the column number of the \(i\)th 1 in row \(j\), converges weakly to the GUE-corners process (that is the joint distribution of the eigenvalues of the left-top parts of a Hermitian random matrix from the Gaussian unitary ensemble). This theorem was conjectured by \textit{V. Gorin} and \textit{G. Panova} [Discrete Mathematics and Theoretical Computer Science. Proceedings, 37--48 (2013; Zbl 1294.05157)] (see also the long write-up on \url{arXiv:1301.0634}).