scaling limit of dependent random walk
Jeonghwa Lee · 2025
Read scaling limit of dependent random walk by Jeonghwa Lee (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-2504-14447.txt.
Introduction
Jeonghwa Lee, writing alone in two thousand twenty-five, finds the large-scale limit of a walk whose steps rest on a fresh kind of linked chance chain. The chain can hold long-range links between its terms.
The limit is a new class of spread paths with no fixed step memory. In it are paths whose links fade at a set rate, as a power, or as a power damped by a set rate. Their spread rules take a split-order form, in some cases damped in time. Two known laws come out as set cases. Lee also treats time-changed forms of the paths, and sets out their rules and link build.
Catalog record
| CREATOR | Jeonghwa Lee |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 88 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2504-14447
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
scaling limit of dependent random walk Thanks: Keywords : Non-Markovian diffusion processes, weak convergence, statistical physics, generalized Bernoulli process, tempered fractional diffusion equation. 2020 MS Classification : 60F17, 60G07, 35D30 Jeonghwa Lee Abstract . Recently, a generalized Bernoulli process (GBP) was developed as a stationary binary sequence that can have long-range dependence. In this paper, we find the scaling limit of a random walk that follows GBP. The result is a new class of non-Markovian diffusion processes. The limiting processes include continuous-time stochastic processes with stationary increments whose correlation decays with an exponential rate, a power law, or an exponentially tempered power law. The limit densities solve a time tempered fractional diffusion equation or time fractional diffusion equation. The second-family of Mittag-Leffler distribution and exponential distribution arise as special cases of the limiting distributions. Subordinated processes are considered as time-changed Lévy processes, and the governing equations and dependence structure of the subordinated processes are discussed. 1. Introduction A random walk describes the path of a walker who takes a random step at discrete, equally distanced time points. The scaling limit of a simple random walk is Brownian motion which was first formulated by [ 8 ] to describe the movements of particles in a liquid. More generally, a Lévy process is infinitely divisible with independent stationary increments, and it is considered a limit of a random walk that takes i.i.d. random steps, which may have an infinite second moment, at equal time intervals [ 9 ] . If a random walk can take large steps with a power-law jump distribution, the resulting Lévy process can model super-diffusion where particles spread faster than the normal diffusion, and its limit density solves a space fractional diffusion equation. In a continuous time random walk (CTRW), random steps are taken at random time points, and it can model both supper- and sub-diffusion [ 17 ] . When the wait times between steps follow a power law distribution, the resulting limiting process is non-Markovian, governed by a (space) time fractional diffusion equation, and it has a wide range of applications in finance, biology, hydrology, etc [ 2 , 3 , 6 , 21 , 22 , 23 ] . Random walks where jumps or wait times follow an exponentially tempered power law distribution can model tempered fractional diffusion [ 1 , 18 , 20 ] . This exponential tempering has both mathematical and practical advantages, and it is proved to be useful in geophysics and finance [ 5 , 20 ] . The tempered fractional Brownian motion is the tempered fractional derivative or integral of a Brownian motion and has stationary increments that can exhibit semi-long range dependence [ 16 ] . More recently, a random walk that takes correlated steps was developed in [ 13 , 15 ] . In [ 15 ] , random steps follow a stationary linear process, which results in the time-changed fractional Brownian motion and linear fractional stable motion as the limiting processes. In [ 13 ] , random steps are obtained by Markov chains through the Bernoulli urn-scheme model and Wright–Fisher model, which results in fractional Pearson diffusion. With a different approach, non-Markovian diffusion processes were developed in [ 19 ] by introducing a memory kernel in the forward drift equation. In this paper, we consider a correlated random walk where steps follow a generalized Bernoulli process (GBP). The GBP was first developed in [ 10 ] as a stationary binary sequence whose covariance function decays with a power law. The sum in a finite sequence of GBP defines the fractional binomial distribution, and its variance is asymptotically proportional to the length of the sequence to a fractional power. In [ 12 ] , it was found that the GBP and the fractional Poisson process have the same large-scale behavior. More specifically, the fractional binomial distribution and the fractional Poisson distribution, when scaled, have the same limiting distribution, which suggests that GBP is a discrete analog of the fractional Poisson process. As the Poisson process can approximate the Bernoulli process under certain conditions, the relationship between the GBP and the fractional Poisson process can be considered its counterpart when there is a long-range dependence. In [ 11 ] , GBP was broadened to include various covariance functions. In this work, we find the scaling limit of a dependent random walk that follows GBP and the dependence structure of the limiting process. The result is a new class of continuous-time stochastic processes with stationary increments whose correlation decays with exponentially tempered power-law. It turns out that the probability density functions of the limiting processes solve time tempered fractional diffusion equations. We also find the limiting process of a random walk that follows a non-stationary GBP, and the limiting distribution includes the second-type Mittag-Leffler distribution and exponentially decaying distribution as special cases. Subordinated processes are considered with a Lévy process whose time is changed by the limiting process of the dependent random walk, which leads to non-Markovian processes that can serve as versatile models for diffusion processes. The marginal density of the resulting processes solves various space-time tempered fractional diffusion equations. Unlike other approaches where the probability density of a diffusion equation was found using an inverse stable subordinator [ 1 ] or forward drift with power kernel [ 19 ] , our method allows us to characterize the full stochastic structure of the processes with its dependence structure, not just the marginal density functions. The paper is organized as follows. In Section 2, we introduce GBP and a dependent random walk that follows GBP. In Section 3, we find the limiting processes of the dependent random walk, providing their moment generating function (mgf) and dependence structure. Subordinated processes are considered as time-changed Lévy processes in Section 4, followed by the conclusion in Section 5. All the proofs are in Section 6. Let us end the introduction with a few words about the notation. We will let i,i_{1},i_{2},i_{3}\cdots\in\mathbb{N}, and i_{1}<i_{2}<i_{3}<\cdots, unless stated otherwise. For x\in\mathbb{R}, \lfloor x\rfloor denotes the largest integer smaller than or equal to x, and [n]=\{1,2,\cdots,n\} for n\in\mathbb{N}. For any two quantities, a,b, that depend on another variable x , a\sim b indicates asymptotic equivalence, i.e., a/b\to 1 as x goes to 0 or \infty. For a set C\subset\mathbb{R}, |C| indicates the number of elements in C , and the ordered elements are denoted by C(1)=\min C,C(2)=\min\{C\setminus\{C(1)\}\}, and in general, the k -th smallest element in C is denoted by C(k) . An indicator function is denoted by I(\cdot) , and generic constants are denoted by c^{\prime} and c^{\prime\prime} that can change from line to line. For notational convenience, we denote P((\cap_{i}A_{i})\bigcap(\cap_{j}B_{j})) by P(\cap_{i}A_{i}\cap_{j}B_{j}). 2. Dependent Random Walk We first develop a generalized Bernoulli process (GBP) similar to [ 10 , 11 ] . A dependent random walk is then defined as a random walk whose steps are dependent and follow GBP. Let \{f_{n}\} be a sequence of functions defined on \mathbb{N}, and b\in(0,1]. Definition 2.1 . Define for a set B=\{i_{0},i_{1},\cdots,i_{k}\}\subset\mathbb{N} with i_{0}<i_{1}<\cdots<i_{k}, L_{n}(B)=\prod_{j=1,\cdots,k}f_{n}(i_{j}-i_{j-1}). If B=\emptyset , define L_{n}(B):=1/(bf_{n}(\lfloor nT\rfloor)) , and if |B|=1,L_{n}(B):=1. Definition 2.2 . Define for disjoint sets, B,F\subset\mathbb{N} with |F|>0, \displaystyle D_{n}(B,F)=\sum_{j=0}^{|F|}\sum_{\begin{subarray}{c}F^{\prime}\subset F\\ |F^{\prime}|=j\end{subarray}}(-1)^{j}L_{n}(B\cup F^{\prime}). If F=\emptyset,D_{n}(B,F):=L_{n}(B). We define GBP in such