Global well-posedness for small data in a 3D temperature-velocity model with Dirichlet boundary noise
Gianmarco Del Sarto, Marta Lenzi · 2025
Read Global well-posedness for small data in a 3D temperature-velocity model with Dirichlet boundary noise by Gianmarco Del Sarto, Marta Lenzi (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-2505-11447.txt.
Introduction
Gianmarco Del Sarto and Marta Lenzi prove that a three-dimensional model of heat and fluid flow, shaken by random noise at its boundary, has exactly one solution when it starts small. Their 2025 paper studies a Boussinesq-type system, where temperature drives motion through buoyancy. Noise enters through the walls as a fixed-temperature condition with a random twist.
The boundary forcing is rough, living in spaces too weak for classical tools. The pair work in a carefully chosen rough space that fits both the flow and the temperature. For small starting data they prove a single mild solution exists up to a stopping time. Better, the chance that the solution lasts the whole interval climbs toward one as the noise shrinks. The result steadies a model class used for heated, stirred fluids.
Catalog record
| CREATOR | Gianmarco Del Sarto, Marta Lenzi |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 107 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2505-11447
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Global well-posedness for small data in a 3D temperature-velocity model with Dirichlet boundary noise Gianmarco Del Sarto, Marta Lenzi Abstract . We study a three-dimensional Boussinesq-type temperature-velocity system on a bounded smooth domain 𝒟 ⊂ ℝ 3 \mathcal{D}\subset\mathbb{R}^{3} , where the velocity u ε u^{\varepsilon} solves the Navier-Stokes equations and the temperature θ ε \theta^{\varepsilon} is driven by Dirichlet boundary noise of intensity ε \sqrt{\varepsilon} . The boundary forcing produces a stochastic convolution Z ε Z^{\varepsilon} which is, in general, only continuous in time with values in H − 1 2 − δ θ ( 𝒟 ) H^{-\frac{1}{2}-\delta_{\theta}}(\mathcal{D}) . To handle this roughness together with initial data θ 0 ∈ W s , 6 / 5 ( 𝒟 ) \theta_{0}\in W^{s,6/5}(\mathcal{D}) , we work in the ambient space H − 1 2 − δ u ( 𝒟 ) H^{-\frac{1}{2}-\delta_{u}}(\mathcal{D}) with δ u ≥ max { δ θ , 1 2 − s } \delta_{u}\geq\max\{\delta_{\theta},\frac{1}{2}-s\} . Given a finite time T > 0 T>0 , for any p > 4 p>4 and sufficiently small initial data, we prove existence and uniqueness of a mild solution ( u ε , θ ε ) (u^{\varepsilon},\theta^{\varepsilon}) up to a stopping time τ ε ≤ T \tau^{\varepsilon}\leq T such that u ε ∈ W 1 , p ( 0 , τ ε , H − 1 2 − δ u ( 𝒟 ) ) ∩ L p ( 0 , τ ε , H 3 2 − δ u ( 𝒟 ) ) , θ ε ∈ C ( 0 , τ ε , H − 1 2 − δ u ( 𝒟 ) ) . u^{\varepsilon}\in W^{1,p}(0,\tau^{\varepsilon};H^{-\frac{1}{2}-\delta_{u}}(\mathcal{D}))\cap L^{p}(0,\tau^{\varepsilon};H^{\frac{3}{2}-\delta_{u}}(\mathcal{D})),\quad\theta^{\varepsilon}\in C(0,\tau^{\varepsilon};H^{-\frac{1}{2}-\delta_{u}}(\mathcal{D})). Moreover, we obtain a high-probability global existence estimate of the form ℙ ( τ ε = T ) ≥ 1 − C ε \mathbb{P}(\tau^{\varepsilon}=T)\geq 1-C\varepsilon , with C = C ( δ θ , T ) > 0 . C=C(\delta_{\theta},T)>0. \labelindent =10pt 1991 Mathematics Subject Classification 60H15, 60H30, 76D03 1. Introduction Let \mathcal{D}\subset\mathbb{R}^{3} be a bounded, open domain (i.e. non-empty, connected set) with smooth boundary \partial\mathcal{D} , and let T>0 be a fixed final time. We study the coupled velocity-temperature system for an incompressible fluid in \mathcal{D} , modelled by Navier-Stokes equations with thermal advection-diffusion, and perturbed by Dirichlet boundary noise. Concretely, for each \varepsilon>0 we consider (1.1) \left\{\begin{aligned} \partial_{t}u^{\varepsilon}+u^{\varepsilon}\cdot\nabla u^{\varepsilon}+\nabla p^{\varepsilon}-\Delta u^{\varepsilon}&=-\theta^{\varepsilon}e_{3},\quad&\text{ in }&\mathcal{D}\times(0,T),\\ \mathrm{div}(u^{\varepsilon})&=0,\quad&\text{ in }&\mathcal{D}\times(0,T),\\ u^{\varepsilon}|_{\partial\mathcal{D}}&=0,\quad&\text{ in }&\partial\mathcal{D}\times(0,T),\\ \partial_{t}\theta^{\varepsilon}+u^{\varepsilon}\cdot\nabla\theta^{\varepsilon}-\Delta\theta^{\varepsilon}&=0,\quad&\text{ in }&\mathcal{D}\times(0,T),\\ \theta_{\mid\partial\mathcal{D}}^{\varepsilon}&=\sqrt{\varepsilon}\,\frac{dW}{dt},\quad&\text{ in }&\partial\mathcal{D}\times(0,T),\\ u^{\varepsilon}|_{t=0}&=u_{0},\quad\theta^{\varepsilon}|_{t=0}=\theta_{0}.\end{aligned}\right. where u^{\varepsilon}\colon\mathcal{D}\times(0,T)\to\mathbb{R}^{3} is the fluid velocity, \theta^{\varepsilon}\colon\mathcal{D}\times(0,T)\to\mathbb{R} is the temperature field, p^{\varepsilon}\colon\mathcal{D}\times(0,T)\to\mathbb{R} is the pressure, e_{3}=(0,0,1)^{T} , and (W_{t})_{t\geq 0} is a \mathcal{Q} -Wiener process acting only on the boundary with intensity which scales depending on the parameter \varepsilon . The parameter \varepsilon influences the probability of global existence up to time T . Indeed, the three-dimensional (3D) Navier-Stokes equations are only known to be globally well - posed under smallness conditions on initial data and forcing (see, e.g., [ 9 , Chapter 9] ). Since stochastic boundary forcing enters the velocity equation through buoyancy coupling, we impose a small - noise regime in order to retain a nonzero probability of well - posedness on [0,T] . However, even for arbitrarily small \varepsilon>0 , the boundary noise may assume large values with positive probability. Consequently, we construct the solution (u^{\varepsilon},\theta^{\varepsilon}) up to a random stopping time \tau^{\varepsilon}\leq T , defined to interrupt the evolution if the stochastic forcing becomes too large. A key difficulty of the model comes from the fact that Dirichlet boundary noise is much rougher than the forcing acting in the interior of the domain. Even in the linear heat equation with boundary noise, the stochastic convolution produced by the boundary forcing typically lives only slightly below the threshold H^{-1/2}(\mathcal{D}) . We quantify this loss through a small exponent \delta_{\theta}>0 , which measures how far the noise-driven component falls into H^{-1/2-\delta_{\theta}}(\mathcal{D}) . To couple this temperature field back into the three-dimensional Navier-Stokes equation via the buoyancy term -\theta^{\varepsilon}e_{3} , we solve the fluid problem in a (slightly) weaker Sobolev scale H^{-1/2-\delta_{u}}(\mathcal{D}) , chosen so that both the noise contribution and the initial temperature can be interpreted as forcing terms at the fluid level. This leads to the compatibility requirement \delta_{u}\geq\max\{\delta_{\theta},\ \tfrac{1}{2}-s\}, where s\in(0,1/2) is the Sobolev regularity of the initial temperature \theta_{0}\in W^{s,6/5}(\mathcal{D}) . Under this condition (and for p>4 in the maximal-regularity framework), we prove the existence and uniqueness of the coupled system on the interval [0,\tau^{\varepsilon}] , satisfying u^{\varepsilon}\in W^{1,p}(0,\tau^{\varepsilon};H^{-\frac{1}{2}-\delta_{u}}(\mathcal{D}))\cap L^{p}(0,\tau^{\varepsilon};H^{\frac{3}{2}-\delta_{u}}(\mathcal{D})),\quad\theta^{\varepsilon}\in C(0,\tau^{\varepsilon};H^{-\frac{1}{2}-\delta_{u}}(\mathcal{D})). Furthermore, we establish the high-probability estimate \mathbb{P}(\tau^{\varepsilon}=T)\geq 1-C\varepsilon . It is worth underlining two aspects of the boundary noise: its mathematical difficulties, and its physical meaning. Firstly, Da Prato and Zabczyk were the first to rigorously demonstrate in [ 10 ] that solutions to the heat equation subject to Dirichlet boundary noise exhibit significantly reduced regularity compared to solutions driven by interior stochastic forcing, even in one spatial dimension. Specifically, due to the presence of boundary noise, the best regularity that one can hope for, even in the linear case, is continuity in time with values in a negative-order Bessel potential space, namely C(0,T;H^{-\frac{1}{2}-\delta_{\theta}}(\mathcal{D})) for any \delta_{\theta}>0 and any final time T>0 . This limitation reflects the difficulties of noisy Dirichlet boundary conditions. The low spatial regularity of the stochastic convolution in the Dirichlet setting is a major obstruction for non-linear problems: even basic products and non-linear maps may fail to be well-defined in a classical sense, and standard energy methods or fixed-point arguments often require additional structure. As a consequence, the available well-posedness theory for non-linear PDEs with Dirichlet white-noise boundary data is rather limited and typically concerns only a limited class of non-linearities; see for instance [ 14 , 24 , 28 ] and references therein. Other works treat situations where the boundary forcing is regularised (e.g. coloured in space and/or fractional in time) or otherwise smoothed at the boundary, which can lead to better regularity and allow one to handle genuinely non-linear dynamics; see [ 1 , 15 ] . In our setting the boundary noise enters the temperature equation linearly, but the resulting rough temperature acts as a forcing in the Navier-Stokes component, and its interplay with the convective non-linearity is one of the main analytic difficulties. It is also worth noting that the H^{-\frac{1}{2}-\delta_{\theta}} limitation is closely related to a boundary-layer singularity: solutions to the associated linear Dirichlet problem may blow up as one approaches \partial\mathcal{D} , while remaining smooth in the interior, see [ 3 , 4 , 8 , 17 ] . In parallel, the velocity field u^{\varepsilon} enjoys, in this context, the maximal regularity of the Stokes operator in the same low-regularity framework introduced above. This constitutes the optimal regularity according to the maximal regularity theory for the Stokes operator on H^{-\frac{1}{2}-\delta_{u}}(\mathcal{D}) , with suitable divergence-free and boundary conditions on the spaces. Our maximal regularity approach follows the framework developed by Prüss and Wilke in [ 27 , Section 5] . Secondly, from a physical point of view, stochastic boundary forcing represents the effect of unresolved fast fluctuations, such as boundary-layer instabilities or small-scale convection, that cannot be described deterministically at the scale of our model. Related approaches that