Chung’s LIL for the linear stochastic fractional heat equation at origin
Chang Liu, Ran Wang · 2025
Read Chung’s LIL for the linear stochastic fractional heat equation at origin by Chang Liu, Ran Wang (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-2511-15228.txt.
Introduction
Liu Chang and Wang Ran establish a fine limit law for a noisy heat equation driven by long-memory noise. The equation spreads an initial state under a fractional Laplacian while a Gaussian noise, white in time and correlated in space like a fractional Brownian motion, keeps stirring it. Laws of the iterated logarithm describe the exact envelope within which such random solutions swing as time grows.
The authors prove Chung's version of that law for the solution at the origin, the sharpest form, which tracks the smallest excursions rather than the largest. The fractional spreading power and the noise memory must sit in stated ranges for the result to hold. Precise envelopes like this anchor the theory of stochastic partial differential equations.
Catalog record
| CREATOR | Chang Liu, Ran Wang |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 25 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2511-15228
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Chung’s LIL for the linear stochastic fractional heat equation at origin Chang Liu, Ran Wang Abstract Abstract: Consider the linear stochastic fractional heat equation with vanishing initial condition: \frac{\partial u(t,x)}{\partial t}=-(-\Delta)^{\frac{\alpha}{2}}u(t,x)+\dot{W}(t,x),\quad t>0,\,x\in\mathbb{R}, where -(-\Delta)^{\frac{\alpha}{2}} denotes the fractional Laplacian with power \alpha\in(1,2] , and the driving noise \dot{W} is a centered Gaussian field which is white in time and has the covariance of a fractional Brownian motion with Hurst parameter H\in\left(\frac{2-\alpha}{2},1\right) . We establish Chung’s law of the iterated logarithm for the solution at t=0 . Keyword: Stochastic fractional heat equation; Chung’s law of the iterated logarithm; Small ball probability. MSC: 60H15; 60G17; 60G22. 1. Introduction Consider the following linear stochastic fractional heat equation (SFHE, for short): \begin{cases}\frac{\partial u(t,x)}{\partial t}=-(-\Delta)^{\frac{\alpha}{2}}u(t,x)+\dot{W}(t,x),\quad t>0,\,x\in\mathbb{R},\\ u(0,x)\equiv 0.\end{cases} (1.1) Here, -(-\Delta)^{\frac{\alpha}{2}} denotes the fractional Laplacian with \alpha\in(1,2] , W(t,x) is a centered Gaussian field with the covariance given by {\mathbb{E}}[W(t,x)W(s,y)]=\frac{1}{2}\left(s\wedge t\right)\left(|x|^{2H}+|y|^{2H}-|x-y|^{2H}\right), (1.2) with \displaystyle\frac{2-\alpha}{2}<H<1. (1.3) That is, W is a random fields, which is a standard Brownian motion in time and a fractional Brownian motion (fBm, for short) with Hurst parameter H in space. According to [ 2 , Theorem 5.3, Remark 5.4] , the equation ( 1.1 ) has a random field solution if and only if ( 1.3 ) holds; See Section 2 for more precise information. In the latter case, the solution \{u(t,x)\}_{t\geq 0,x\in\mathbb{R}} can be written as u(t,x)=\int_{0}^{t}\int_{\mathbb{R}}G(t-s,x-y)W(ds,dy),\ \ \text{a.s.} (1.4) where G(t,x) is the Green kernel associated with the operator -(-\triangle)^{\frac{\alpha}{2}} is given by (\mathcal{F}G(t,\cdot))(\xi)=e^{-t|\xi|^{\alpha}},\ \ \xi\in{\mathbb{R}},\ \ t>0. (1.5) Using the idea of the pinned string process from Mueller and Tribe [ 17 ] , it can be shown that the solution to the linear stochastic (fractional) heat equation can be decomposed into the sum of two Gaussian random fields: one possessing stationary increments and the other exhibiting smooth sample paths. See [ 5 , 6 , 13 , 8 , 20 , 23 ] , as well as the monographs [ 4 , 9 , 19 ] . For example, when \dot{W} is a space-time white noise (i.e., H=1/2 ), Theorem 3.3 in [ 9 ] tells us that for any fixed x\in\mathbb{R} , the stochastic process t\mapsto u(t,x) can be represented as the sum of a fBm with Hurst parameter 1/4 and a stochastic process with C^{\infty}(0,\infty) -continuous trajectories. Hence, locally t\mapsto u(t,x) behaves as a fBm with Hurst parameter 1/4 , and it has the same regularity (such as the Hölder continuity, the local moduli of continuities, Chung-type law of the iterated logarithm) as a fBm with Hurst parameter 1/4 . For the solution of the equation ( 1.1 ), according to [ 5 , Section 3.2] , we know that for any fixed x\in\mathbb{R} , there exists a Gaussian process \{T(t)\}_{t\geq 0} , which is independent of \{u(t,x)\}_{t\geq 0} and has a version that is infinitely differentiable on (0,\infty) , such that \displaystyle F:=\kappa^{-1}\left\{u(t,x)+T(t)\right\}_{t\geq 0} (1.6) is a fBm with Hurst parameter \theta , defined by \theta:=\frac{1}{2}-\frac{1-H}{\alpha}. (1.7) Here, the constant \kappa is given by \kappa:=\left(\frac{H\Gamma(2H)\Gamma(1-2\theta)\sin(\pi H)}{\pi\alpha\theta}\right)^{\frac{1}{2}}. (1.8) The decomposition of ( 1.6 ), together with Chung’s LIL of the fBm (e.g., [ 14 , Section] , [ 18 ] ), implies that for every fixed t\in(0,\infty) and x\in\mathbb{R} , \liminf_{\varepsilon\downarrow 0}\sup_{h\in[0,\varepsilon]}\frac{|u(t+h,x)-u(t,x)|}{{\varepsilon}^{\theta}(\log\log{\varepsilon}^{-1})^{-\theta}}=\kappa\lambda^{\theta},\ \ \ a.s., (1.9) where \lambda is the small ball constant of a fractional Brownian motion with index \theta (e.g. [ 14 , Theorem 6.9] ). However, the argument in the proof of Chung’s LIL of u(t,x) at t>0 does not work in the case of t=0 and the problem on Chung’s LIL for u(t,x) at t=0 was left open in [ 5 ] . This effort is complicated by the fact that, near t=0 , the random process T is not smooth; in fact, T and F are equally smooth locally near t=0 . When \alpha=2 and H=\frac{1}{2} , Khoshnevisan, Kim, and Mueller [ 10 ] recently proved that the small-ball probability of the rougher process F to dominate that of the smoother Gaussian process T , and they adopted a localization idea of Lee and Xiao [ 12 ] to establish Chung’s LIL of u(t,x) at t=0 . The objective of this paper is to establish Chung’s LIL at t=0 for the solution u(t,x) of the linear stochastic fractional heat equation ( 1.1 ), for any fixed x\in\mathbb{R} , with parameters \alpha\in(1,2] and H\in(1-\alpha/2,1) . Theorem 1.1 . For any fixed x\in\mathbb{R} , \liminf_{\varepsilon\downarrow 0}\sup_{t\in[0,\varepsilon]}\frac{|u(t,x)|}{{\varepsilon}^{\theta}(\log\log{\varepsilon}^{-1})^{-\theta}}=\kappa\lambda^{\theta},\ \ \ a.s., (1.10) where \lambda is the small ball constant of a fractional Brownian motion with index \theta . The lower bound in ( 1.10 ) is derived from a small ball probability result (Lemma 2.1 ), following the classical method of Monrad and Rootzén [ 16 ] and Talagrand [ 18 ] . The upper bound, however, requires a more intricate analysis. To establish this upper bound, we adapt the localization approach introduced by Lee and Xiao [ 12 ] and Khoshnevisan et al. [ 10 ] . The paper is structured as follows. In Section 2, we recall some properties of the solution, and Borell’s inequality and Dudley’s metric entropy theorem. In Section 3, we prove Theorem 1.1 by analyzing the qualities of the approximation errors. 2. Preliminaries 2.1. Properties of the solution Let us recall some notations from [ 7 ] . Let \mathcal{D}=\mathcal{D}(\mathbb{R}) be the space of real-valued infinitely differentiable functions with compact support on \mathbb{R} . The Fourier transform of a function f\in\mathcal{D} is defined as \mathcal{F}f(\xi):=\int_{\mathbb{R}}e^{-i\xi x}f(x)dx. Let \mathcal{D}(\mathbb{R}_{+}\times\mathbb{R}) denote the space of real-valued infinitely differentiable functions with compact support in \mathbb{R}_{+}\times\mathbb{R} and let W be a zero-mean Gaussian family \{W(\phi),\phi\in\mathcal{D}(\mathbb{R}_{+}\times\mathbb{R})\} , whose covariance structure is given by \mathbb{E}\big[W(\phi)W(\psi)\big]=C_{H}\int_{\mathbb{R}_{+}\times\mathbb{R}}\mathcal{F}\phi(s,\xi)\overline{\mathcal{F}\psi(s,\xi)}\cdot|\xi|^{1-2H}d\xi ds, (2.1) where H\in(0,1) , C_{H}=\frac{1}{2\pi}\Gamma(2H+1)\sin(\pi H) and \mathcal{F}\phi(s,\xi) is the Fourier transform of \phi(s,x) with respect to the spatial variable x . Let \mathcal{H} be the completion of \mathcal{D}(\mathbb{R}_{+}\times\mathbb{R}) with the inner product \langle\phi,\psi\rangle_{\mathcal{H}}:=\mathbb{E}\big[W(\phi)W(\psi)\big]. The mapping \varphi\mapsto W(\varphi) can be extended to all \varphi\in\mathcal{H} . According to Balan [ 2 , Remark 5.4] and [ 15 , Proposition 2.2] , we have the following results about the solution to Equation ( 1.1 ). Proposition 2.1 . Assume \alpha\in(1,2] and \frac{2-\alpha}{2}<H<1 . Then, the equation ( 1.1 ) admits a unique solution \{u(t,x);(t,x)\in\mathbb{R}_{+}\times\mathbb{R}\} . Moreover, the following items hold. (a) For any t\geq 0 and x\in\mathbb{R} , \|u(t,x)\|_{L^{2}(\Omega)}=\sqrt{c_{2,1}}t^{\theta}, (2.2) where c_{2,1}=\frac{C_{H}}{2H+\alpha-2}2^{\frac{2H+\alpha-2}{\alpha}}\Gamma\left(\frac{2-2H}{\alpha}\right) . (b) For any fixed x\in\mathbb{R} , there exists c_{2,2}>0 such that for any t,s\in\mathbb{R}_{+} , \|u(t,x)-u(s,x)\|_{L^{2}(\Omega)}\leq c_{2,2}|t-s|^{\theta}. (2.3) (c) For any fixed x\in\mathbb{R} , the process \{u(t,x)\}_{t\geq 0} is a Gaussian self-similar process with index \theta given by ( 1.7 ). Proof. The results in (a) and (b) are taken from [ 15 , Proposition 2.2, Lemma 2.1] . Now, we give the proof for (c). For any fixed x\in\mathbb{R} , \rho>0 and 0<s<t , by ( 2.1 ) and the change of variables, we have \begin{split}&{\mathbb{E}}\left[\rho^{-\theta}u\left(\rho t,x\right)\rho^{-\theta}u\left(\rho s,x\right)\right]\\ =&\,C_{H}\rho^{-2\theta}\int_{0}^{\rho s}\int_{\mathbb{R}}\mathcal{F}G\left(\rho t-r,x-\cdot\right)(\xi)\overline{\mathcal{F}G\left(\rho s-r,x-\cdot\right)(\xi)}|\xi|^{1-2H}d\xi dr\\ =&\,C_{H}\rho^{-2\theta}\int_{0}^{\rho s}\int_{\mathbb{R}}e^{-(\rho t+\rho s-2r)|\xi|^{\alpha}}|\xi|^{1-2H}d\xi dr\\ =&\,C_{H}\int_{0}^{s}\int_{\mathbb{R}}e^{-(t+s-2r)|\xi|^{\alpha}}|\xi|^{1-2H}d\xi dr\\ =&\,{\mathbb{E}}\left[u(t,x)u(s,x)\right].\end{split} (2.4) The proof is complete. ∎ Let us recall a small ball estimate result for the process \{u(t,x)\}_{t\geq 0} for any fixed x\in\mathbb{R} . Lemma 2.1 . [ 5 , Theorem 4.3] For every fixed x\in\mathbb{R} , \lim\limits_{\varepsilon\downarrow 0}\varepsilon^{\frac{1}{\theta}}\log\mathbb{P}\left(\sup_{t\in[0,1]}|u(t,x)|\leq\varepsilon\right)=-\kappa^{\frac{1}{\theta}}\lambda, (2.5) where \lambda is the small ball constant