Random vicious walks and random matrices (Q2711839)
From MaRDI portal
| This is the item page for this Wikibase entity, intended for internal use and editing purposes. Please use this page instead for the normal view: scientific article |
scientific article
| Language | Label | Description | Also known as |
|---|---|---|---|
| English | Random vicious walks and random matrices |
scientific article |
Statements
26 April 2001
0 references
vicious walkers
0 references
lock step walk
0 references
totally asymmetric exclusion process
0 references
Tracy-Widom distribution
0 references
Gaussian orthogonal ensemble
0 references
0 references
0.9500097
0 references
0.9241028
0 references
0 references
0.8990791
0 references
0.89204466
0 references
0.89163476
0 references
Random vicious walks and random matrices (English)
0 references
The author considers an infinite system of nearest-neighbor walks on \(\mathbb Z\) in discrete time. At time zero, there is one walker each at the points \(0, 2, 4, 6,\dots\). At each discrete time unit \(1,2,3,\dots\), every walker makes a step with equal probability to the right or left. The only condition is that no two walkers may be at one site at the same time, i.e., the walkers are non-colliding. Related walker systems appear in the literature under the notion random turn walks or lock step walks. The model is equivalent to a certain totally asymmetric exclusion process. NEWLINENEWLINENEWLINEThe main result of the paper is the identification of the limiting distribution of the number of left steps made by the leftmost walker by time \(N\), denoted \(L_1(N)\), conditioned on the event \(E(N,k)\) that in the system a total of \(k\) steps has been made up to time \(N\). Here \(k\) is coupled with \(N\) as \(k=t^2 N^2/(1-t^2)+o(N^{4/3})\), where \(t\in(0,1)\) is a parameter. It turns out that the limiting distribution is an elementary transformation of the well-known GOE Tracy-Widom distribution, which appeared for the first time in 1994 as the (renormalized and rescaled) limiting distribution of the largest eigenvalue of the Gaussian orthogonal ensemble, GOE. More precisely, denote this distribution function by \(F_1\) (it is defined in terms of a solution of the Painlevé II equation), then the author obtains that NEWLINE\[NEWLINE \lim_{N\to\infty} \mathbb P\Bigl(\frac{L_1(N)-\eta(t)N}{\rho(t) N^{1/3}} \leq x \Big|E(N,k)\Bigr)=F_1(x),\qquad x\in\mathbb R, NEWLINE\]NEWLINE where \(\eta(t)=2t/(1+t)\) and \(\rho(t)=(1(1-t))^{1/3}/(1+t)\). Furthermore, it is also proved that all the rescaled conditional moments of \(L_1(N)\) converge to the corresponding moments of the Tracy-Widom distribution. Extensions to the \(j\)th walker for any \(j\in \mathbb N\) are discussed and contrasted with the corresponding results for the random turn walk. The main tools are a bijection between certain path system classes and classes of semistandard Young tableaux [established by \textit{A. J. Guttmann, A. L. Owczarek} and \textit{X. G. Viennot}, J. Phys. A, Math. Gen. 31, 8123-8135 (1998; Zbl 0930.05098)], representations of the generating function for the first row in terms of Hankel determinants, orthogonal polynomials, a Riemann-Hilbert approach and the Deift-Zhou steepest-descent method.
0 references