On the passage times of self-similar Gaussian processes on curved boundaries
Davar Khoshnevisan and Cheuk Yin Lee · 2025
Read On the passage times of self-similar Gaussian processes on curved boundaries by Davar Khoshnevisan and Cheuk Yin 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-2506-15949.txt.
Introduction
Davar Khoshnevisan and Cheuk Yin Lee pin how long a self-like chance path takes to first cross a bent edge. The edge grows as a set power of time; the path scales the same at all zooms, as in the stock walk's kin. Their 2025 proof sorts three worlds by how the edge power stands next to the path's own scale mark. Past the scale, the cross may never come.
Short of it, with a strong spread trait in the path, the wait holds all span means. At the tie, a drop curve splits waits with sound means from waits with none. The proof grows a famed old law for the plain walk at its square-root edge. Two case sets show the craft at work.
Catalog record
| CREATOR | Davar Khoshnevisan and Cheuk Yin Lee |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 70 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2506-15949
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
On the passage times of self-similar Gaussian processes on curved boundaries Thanks: Research supported in part by the National Science Foundation grant DMS-2245242 and research startup fund of the Chinese University of Hong Kong, Shenzhen Davar Khoshnevisan and Cheuk Yin Lee Abstract . Let T c , β T_{c,\beta} denote the smallest t ⩾ 1 t\geqslant 1 that a continuous, self-similar Gaussian process with self-similarity index α > 0 \alpha>0 moves at least ± c t β \pm ct^{\beta} units. We prove that: (i) If β > α \beta>\alpha , then T c , β = ∞ T_{c,\beta}=\infty with positive probability; (ii) If β < α \beta<\alpha then T c , β T_{c,\beta} has moments of all order; and (iii) If β = α \beta=\alpha and X X is strongly locally nondeterministic in the sense of Pitt [ 88 ] , then there exists a continuous, strictly decreasing function λ : ( 0 , ∞ ) → ( 0 , ∞ ) \lambda:(0\,,\infty)\to(0\,,\infty) such that E ( T c , β μ ) \mathrm{E}(T_{c,\beta}^{\mu}) is finite when 0 < μ < λ ( c ) 0<\mu<\lambda(c) and infinite when μ > λ ( c ) \mu>\lambda(c) . Together these results extend a celebrated theorem of Breiman [ 14 ] and Shepp [ 94 ] for passage times of a Brownian motion on the critical square-root boundary. We briefly discuss two examples: One about fractional Brownian motion, and another about a family of linear stochastic partial differential equations. 1991 Mathematics Subject Classification 60G15; 60G18 1. Introduction Choose and fix a number c>0 , let w=\{w_{n}\}_{n=1}^{\infty} denote the simple symmetric random walk on \mathbb{Z} , and define \tau_{c}=\inf\{n\in\mathbb{N}:\,|w_{n}|>c\sqrt{n},\ c\sqrt{n}>1\} , where \inf\varnothing=\infty . It is not hard to see that \mathrm{P}\{\tau_{c}<\infty\}=1 for every c>0 . This follows for example from the law of the iterated logarithm (LIL). Blackwell and Freedman [ 11 ] proved that (1.1) \mathrm{E}(\tau_{c})<\infty\quad\text{if and only if}\quad c<1. Moreover, w can be replaced for example by any random walk that satisfies \mathrm{E}(w_{1})=0 and \Var(w_{1})=1 ; see Gundy and Siegmund [ 47 ] . This type of result has been studied in the continuum as well. Let W denote a standard 1-D Brownian motion, and define T_{c}=\inf\left\{t\geqslant 1:\,|W(t)|>ct^{1/2}\right\}\qquad\forall c>0. By the LIL, T_{c}<\infty a.s. for every c>0 . Breiman [ 14 ] and Shepp [ 94 ] proved at the same time that, for every c,\mu>0 , \mathrm{E}(T_{c}^{\mu})<\infty\quad\text{if and only if}\quad c<\mathcal{z}(\mu), where \mathcal{z}(\mu) denotes the smallest positive root of x\mapsto M(-\mu\,,1/2\,,x^{2}/2) where M denotes Kummer’s function. The function \mu\mapsto\mathcal{z}(\mu) cannot be evaluated explicitly but it is known to have a continuous inverse c\mapsto\mathcal{z}^{-1}(c) , and (1.2) \begin{split}\mathcal{z}(\mu)=\frac{1+\mathcal{o}(1)}{\sqrt{\mu}}\text{ as }\mu\to\infty,\quad&\mathcal{z}(\mu)=\sqrt{(2+\mathcal{o}(1))|\log\mu|}\text{ as }\mu\to 0^{+},\\ \mathcal{z}^{-1}(c)=\frac{1+\mathcal{o}(1)}{c^{2}}\text{ as }c\to 0^{+},\quad&\mathcal{z}^{-1}(c)=\exp\left(-c^{2}\left[\tfrac{1}{2}+\mathcal{o}(1)\right]\right)\text{ as }c\to\infty.\end{split} Moreover, some special evaluations of \mathcal{z} are also known exactly [ 14 , 94 ] ; for instance, (1.3) \textstyle\mathcal{z}(1)=1\quad\text{and}\quad\mathcal{z}(2)=\sqrt{3-\sqrt{6}}. Novikov [ 82 ] presents exact formulas for closely related passage times. And Breiman ( loc. cit. ) also derived an invariance principle for a large class of random walks in the domain of attraction of W that include the simple walk. In particular, Breiman’s theory implies that \mathrm{E}(\tau_{c}^{\mu})<\infty if and only if c<\mathcal{z}(\mu). Breiman’s theorem includes ( 1.1 ) thanks to ( 1.3 ). It also includes the fact that \mathrm{E}(\tau_{c}^{2})<\infty if and only if c<\sqrt{3-\sqrt{6}}\approx 0.742, established originally by Chow and Teicher [ 19 ] . See also Teicher and Zhang [ 99 ] together with their detailed bibliography to the previous work on this topic. Alili and Patie [ 1 ] discuss some of the latest developments of passage times of Brownian motion on curved boundaries. In addition to their consequeneces in theoretical statistics, passage-time problems are also intimately connected to the foundations of martingale theory. Notably, Davis [ 25 ] found the optimal constant in the celebrated Burkholder-Davis-Gundy inequality in part by establishing the extremality of the stopping times \{T_{c}\}_{c>0} . Finally, we mention that such passage times also play a vital role in the deep analysis of slow points of the increments of Brownian motion and related processes [ 4 , 26 , 27 , 44 , 43 , 87 , 101 ] . In this paper, we aim to study the passage times of self-similar Gaussian processes on curved boundaries. With this in mind, from now on we let X=\{X(t)\}_{t\geqslant 0} denote a centered Gaussian process with continuous sample functions, and we assume throughout that X is \alpha -self similar for some number \alpha>0 whose value is fixed from now on. Throughout, X is assumed also to be non degenerate; that is, \Var[X(1)]>0 . It might help to recall that \alpha -self similarity means that the law of \{r^{-\alpha}X(rt)\}_{t\geqslant 0} does not depend on r>0 [ 56 , 72 ] . This notion was studied in depth by Lamperti [ 61 , 62 , 63 ] who alternatively referred to self-similar processes as “semi stable.” For Markov processes, the subject of self-similarity has been revisited more recently with remarkable advances. The bibliography has pointers to portions of that literature that are somewhat related to the topic of the present work [ 3 , 8 , 9 , 10 , 15 , 18 , 16 , 17 , 29 , 28 , 31 , 32 , 40 , 42 , 55 , 57 , 68 , 76 , 84 , 85 , 86 , 89 , 90 , 91 , 102 , 105 ] . Self-similar Gaussian processes have also been the subject of many impactful papers a representative collection of which can also be found in the references [ 21 , 20 , 30 , 39 , 41 , 46 , 51 , 49 , 50 , 52 , 53 , 70 , 73 , 74 , 78 , 81 , 83 , 95 , 96 , 107 ] . Recall that X is strongly locally non-deterministic (SLND, for short) [ 5 , 6 , 79 , 22 , 104 , 88 ] if and only if there exists a number {l}_{{}_{X}}>0 such that \Var\left(X(t)\mid X(r);\,0<r<s\right)\geqslant{l}_{{}_{X}}(t-s)^{2\alpha}, uniformly for all 0<s<t . Strong local non-determinism is not always assumed. Whenever we assume that X satisfies ( 1 ) we explicitly mention that assumption. The following summarizes what we currently know about the passage times of self-similar, continuous Gaussian processes on curved boundaries of the form ct^{\beta} , and can be viewed as a nontrivial generalization of the results of Breiman [ 14 ] and Shepp [ 94 ] , valid in the absence of both the theories of Markov processes and martingales. Consider the following stopping times of the Gaussian process X : T_{c,\beta}=\inf\{t\geqslant 1:\ |X(t)|>ct^{\beta}\}\qquad\forall c,\beta>0. The following is the main result of this paper. The super- and the sub-critical cases of the next result [parts (1) and (2)] are related to recent testing methodologies that are proposed in clinical trials [ 59 , 60 , 106 ] when X denotes a fractional Brownian motion. Theorem 1.1 . (1) If \beta>\alpha , then 0<\mathrm{P}\{T_{c,\beta}=\infty\}<1 for all c>0 ; (2) If 0<\beta<\alpha , then (1.4) \mathrm{E}\left({\rm e}^{A|\log T_{c,\beta}|^{2}}\right)<\infty, for all c>0 and all sufficiently small A=A(c\,,\beta)>0 . Consequently, \mathrm{E}(T_{c,\beta}^{\mu})<\infty for all c,\mu>0 in this case; (3) For every c>0 there exists a