Boundary Feller-Dynkin processes associated with Laguerre processes and Pickrell diffusions
ALEXANDER I. BUFETOV, YOSUKE KAWAMOTO · 2025
Read Boundary Feller-Dynkin processes associated with Laguerre processes and Pickrell diffusions by ALEXANDER I. BUFETOV, YOSUKE KAWAMOTO (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2509-17045.txt.
Introduction
Alexander I. Bufetov and Yosuke Kawamoto wrote a 2025 arXiv paper in pure mathematics about boundary processes tied to Laguerre processes. The work builds on an earlier study by the same pair, where they set up a linked family of these processes with one fixed parameter. Here they ask what process appears at the boundary of that family.
To answer, the authors use a method built on operators that link one process to the next. They work out the boundary process in full, and it turns out to be a system that moves in a fixed, rule-based way with no randomness left. The same frame also fits Pickrell diffusions, so the paper derives the random boundary process for that family as well.
Catalog record
| CREATOR | ALEXANDER I. BUFETOV, YOSUKE KAWAMOTO |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 80 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2509-17045
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Boundary Feller-Dynkin processes associated with Laguerre processes and Pickrell diffusions ALEXANDER I. BUFETOV, YOSUKE KAWAMOTO Abstract . In this paper, we construct a Feller-Dynkin boundary process by applying the method of intertwiners to the coherent family, introduced in our previous work, of Laguerre processes with a fixed parameter. The corresponding boundary process is computed explicitly, and it turns out to be a deterministic dynamical system. Since the Pickrell diffusions that leave the Pickrell measure invariant are coherent with respect to the same projective system, we also obtain the stochastic boundary process associated with the coherent family of Pickrell diffusions. 1991 Mathematics Subject Classification 60B20, 60J60 To Yakov Grigorievich Sinai on the occasion of his 90th birthday 1. Introduction 1.1. The method of intertwiners The aim of this paper is to give an explicit construction of boundary Feller-Dynkin processes associated with the projective system introduced in [ 12 ] , by applying the method of intertwiners. Infinite-dimensional dynamics can be constructed in many examples [ 7 , 8 , 29 , 13 , 3 ] , and the boundary process is sometimes a stochastic infinite-dimensional diffusion [ 28 ] . Sometimes, however, a surprising phenomenon occurs, and in transition to infinite dimension the stochasticity disappears: the limit diffusion turns out to be a deterministic dynamical system [ 3 ] . In this paper, we construct a projective system and verify that the resulting infinite-dimensional system is indeed deterministic, see ( 1.6 ) below. We begin with a brief overview of this approach. A detailed rigorous formulation is provided in Section 2 . A projective system \{E^{N},L_{N}^{N+1}\}_{N\in\mathbb{N}} consists of a sequence of spaces E^{N} and Markov kernels L_{N}^{N+1} from E^{N+1} to E^{N} . The boundary of a projective system is again a pair consisting of a space and a Markov kernel. The boundary E^{\infty} is a projective limit of the spaces \{E^{N}\}_{N\in\mathbb{N}} in a suitable sense. Let \{T_{t}^{N}\}_{t\geq 0} be the Markov semigroup associated with a Markov process \mathbf{X}^{N} on E^{N} . If the intertwining relation \displaystyle T_{t}^{N+1}L_{N}^{N+1}=L_{N}^{N+1}T_{t}^{N}\text{ for any }t\geq 0 holds for each N\in\mathbb{N} , then we say that the family of Markov processes \{\mathbf{X}^{N}\}_{N\in\mathbb{N}} is coherent (or consistent) with respect to \{E^{N},L_{N}^{N+1}\}_{N\in\mathbb{N}} . Under certain additional conditions, a coherent family \{\mathbf{X}^{N}\}_{N\in\mathbb{N}} induces a corresponding Markov process \mathbf{X}^{\infty} on the boundary E^{\infty} . Furthermore, if the kernels L_{N}^{N+1} are Feller and the Markov processes \{\mathbf{X}^{N}\}_{N\in\mathbb{N}} are Feller-Dynkin, then the boundary process X^{\infty} is also Feller-Dynkin. The method of intertwiners was introduced by Borodin and Olshanski [ 7 ] . They constructed a Feller-Dynkin process on the boundary of the Gelfand-Tsetlin graph, which describes the branching of irreducible representations of the chain of unitary groups. This approach was further applied to branching graphs associated with other groups by Borodin and Olshanski [ 8 , 29 ] and by Cuenca [ 13 ] . In their frameworks, the state spaces of the projective systems are discrete. Assiotis [ 3 ] first applied the method of intertwiners to a continuous setting. Let us explain the framework by Assiotis. Let \mathring{W}^{N}=\{\mathbf{x}=(x_{i})_{i=1}^{N}\in\mathbb{R}^{N}\,;\,x_{1}<\cdots<x_{N}\} be the Weyl chamber and W^{N}=\{\mathbf{x}\in\mathbb{R}^{N}\,;\,x_{1}\leq\cdots\leq x_{N}\} be its closure. For \mathbf{x}=(x_{i})_{i=1}^{N+1}\in W^{N+1} , we introduce the set \displaystyle W^{N,N+1}(\mathbf{x}) \displaystyle=\{\mathbf{y}\in W^{N}\,;\,x_{1}\leq y_{1}\leq x_{2}\leq\cdots\leq y_{N}\leq x_{N+1}\} For \mathbf{y}=(y_{i})_{i=1}^{N}\in W^{N} , denote \displaystyle\Delta_{N}(\mathbf{y})=\prod_{1\leq i<j\leq N}(y_{j}-y_{i}), the Vandermonde determinant. In the particular case \mathbf{x}\in\mathring{W}^{N+1} , we define the probability measure \Lambda_{N}^{N+1}(\mathbf{x},\cdot) on W^{N} by the formula (1.1) \displaystyle\Lambda_{N}^{N+1}(\mathbf{x},d\mathbf{y})=N!\cdot\frac{\Delta_{N}(\mathbf{y})}{\Delta_{N+1}(\mathbf{x})}\mathbf{1}_{W^{N,N+1}(\mathbf{x})}(\mathbf{y})d\mathbf{y}. The kernel \Lambda_{N}^{N+1} can be extended to a Feller kernel W^{N+1}\dashrightarrow W^{N} [ 3 , Lemma 2.5] . Thus, we have a projective system \{W^{N},\Lambda_{N}^{N+1}\}_{N\in\mathbb{N}} . The Dyson Brownian motions for the inverse temperature \beta=2 , and its Ornstein-Uhlenbeck counterparts, form consistent families with respect to this projective system [ 4 , Section 3.8] , [ 35 , Section 3] (see also [ 17 , 32 ] for general \beta ). Furthermore, the family of Markov processes leaving the Hua-Pickrell measures invariant is also coherent [ 3 ] . Therefore, these coherent systems give rise to the associated boundary processes by a framework established in [ 3 ] . The Laguerre processes are also intertwined by \Lambda_{N}^{N+1} [ 1 , 4 ] , but in this case, the parameter of the process varies with N (shifted intertwining relation). As a result, the Laguerre processes with fixed parameter are not coherent families with respect to \{W^{N},\Lambda_{N}^{N+1}\}_{N\in\mathbb{N}} . In [ 12 ] , we introduced a new kernel \Lambda_{\alpha,N}^{N+1} by which the Laguerre processes with fixed parameter are intertwined. Let W_{\geq}^{N}=W^{N}\cap[0,\infty)^{N} denote the closure of the non-negative Weyl chamber. Let \alpha>-1 be a real number. For \mathbf{x}\in\mathring{W}_{\geq}^{N+1}:=\{\mathbf{x}\in\mathbb{R}^{N+1}\,;\,0<x_{1}<\cdots<x_{N+1}\} , we define \Lambda_{\alpha,N}^{N+1}(\mathbf{x},\cdot) as a probability measure on W_{\geq}^{N} given by (1.2) \displaystyle\Lambda_{\alpha,N}^{N+1}(\mathbf{x},d\mathbf{y})=N!(\alpha+1)_{N}\frac{\Delta_{N}(\mathbf{y})}{\Delta_{N+1}(\mathbf{x})}\prod_{k=1}^{N}\bigg(\mathbf{1}_{[x_{k-1},x_{k+1}]}(y_{k})\int_{x_{k}\vee y_{k}}^{x_{k+1}\wedge y_{k+1}}\frac{y_{k}^{\alpha}}{z^{\alpha+1}}dz\bigg)d\mathbf{y}. Here, we use the symbols x_{0}=0 and y_{N+1}=\infty for notational convenience. We also use the symbol (x)_{n}=x(x+1)\cdots(x+n-1) to denote the shifted factorial. Since the definition of ( 1.2 ) is valid for \mathbf{x}\in\mathring{W}_{\geq}^{N+1} , the kernel \Lambda_{\alpha,N}^{N+1} is extended to a Feller kernel W_{\geq}^{N+1}\dashrightarrow W_{\geq}^{N} [ 12 , Lemma 5, Proposition 8] . In this paper, we apply the method of intertwiners to the projective system \{W_{\geq}^{N},\Lambda_{\alpha,N}^{N+1}\}_{N\in\mathbb{N}} for \alpha\in\{0\}\cup\mathbb{N} . As a result, the boundary process corresponding to the Laguerre processes are obtained. The boundary process turns out to be deterministic, and we find it explicitly: see ( 1.6 ) below. Furthermore, we introduce diffusion processes leaving the Pickrell ensemble, the radial parts distribution of the Pickrell measure, invariant. These dynamics are also intertwined by \Lambda_{\alpha,N}^{N+1} , and determines a boundary processes. To state our main results, we give an informal description of a boundary of the projective system. A precise definition will be given after the main results. A boundary of the projective system \{W_{\geq}^{N},\Lambda_{\alpha,N}^{N+1}\}_{N\in\mathbb{N}} is given by (1.3) \displaystyle\Omega=\{(\boldsymbol{\alpha},\gamma)\,;\,\boldsymbol{\alpha}=(\alpha_{i})_{i\in\mathbb{N}},\alpha_{1}\geq\alpha_{2}\geq\cdots\geq 0,\,\sum_{i\in\mathbb{N}}\alpha_{i}\leq\gamma\}\subset\mathbb{R}^{\mathbb{N}}\times\mathbb{R} endowed with product topology. By the definition of boundary, there exists a kernel \Lambda_{\alpha,N}^{\Omega}:\Omega\dashrightarrow W_{\geq}^{N} such that, for each N\in\mathbb{N} , we have \displaystyle\Lambda_{\alpha,N+1}^{\Omega}\Lambda_{\alpha,N}^{N+1}=\Lambda_{\alpha,N}^{\Omega}. 1.2. The boundary Laguerre process We first state our result on the Laguerre processes. For \alpha>-1 , we consider the N -dimensional stochastic differential equation (1.4) \displaystyle dX_{t}^{N,i} \displaystyle=\sqrt{2X_{t}^{N,i}}dB_{t}^{i}+\Big(-X_{t}^{N,i}+\alpha+N+\sum_{j\neq i}^{N}\frac{X_{t}^{N,i}+X_{t}^{N,j}}{X_{t}^{N,i}-X_{t}^{N,j}}\Big)dt for i=1,\ldots,N . Here, (B^{i})_{i=1}^{N} denotes the standard N -dimensional Brownian motion. For \alpha>-1 , the equation ( 1.4 ) has a unique strong solution for any starting point \mathbf{x}\in W_{\geq}^{N} [ 19 , Theorem 2.2] . We call the unique strong solution \mathbf{X}^{N}=(X^{N,1},\ldots,X^{N,N}) the Laguerre (Ornstein-Uhlenbeck) process of parameter \alpha . Let \{T_{\alpha,t}^{N}\}_{t\geq 0} be the Markov semigroup associated with \mathbf{X}^{N} . Note that the equation ( 1.4 ) is the Ornstein-Uhlenbeck analogue of the non-colliding squared Bessel processes, which have been extensively studied in the context of random matrix theory (see [ 21 ] for example). In fact, the non-colliding squared Bessel process describes the evolution of the squared singular value of a rectangular matrix whose entries are independent complex Brownian motions [ 15 , 25 ] (see also [ 9 ] for the real-valued case). Define C_{\infty}(W_{\geq}^{N}) as the space of continuous functions on W_{\geq}^{N} vanishing at infinity. We established the following intertwining relation. Proposition 1.1 . [ 12 , Theorem 1] Assume that \alpha>-1 . Then, for any N\in\mathbb{N} , f\in C_{\infty}(W_{\geq}^{N}) , and t\geq 0 , we have the intertwining relation (1.5) \displaystyle T_{\alpha,t}^{N+1}\Lambda_{\alpha,N}^{N+1}f \displaystyle=\Lambda_{\alpha,N}^{N+1}T_{\alpha,t}^{N}f. Remark 1.2 . The intertwining relation ( 1.5 ) corresponds