Regularization by rough Kraichnan noise for the generalised SQG equations

Marco Bagnara,Lucio Galeati,Mario Maurelli
2024-07-26
Abstract:We consider the generalised Surface Quasi-Geostrophic (gSQG) equations in $\mathbb R^2$ with parameter $\beta\in (0,1)$, an active scalar model interpolating between SQG ($\beta=1$) and the 2D Euler equations ($\beta=0$) in vorticity form. Existence of weak $(L^1\cap L^p)$-valued solutions in the deterministic setting is known, but their uniqueness is open. We show that the addition of a rough Stratonovich transport noise of Kraichnan type regularizes the PDE, providing strong existence and pathwise uniqueness of solutions for initial data $\theta_0\in L^1\cap L^p$, for suitable values $p\in[2,\infty]$ related to the regularity degree $\alpha$ of the noise and the singularity degree $\beta$ of the velocity field; in particular, we can cover any $\beta\in (0,1)$ for suitable $\alpha$ and $p$ and we can reach a suitable ("critical") threshold. The result also holds in the presence of external forcing $f\in L^1_t (L^1\cap L^p)$ and solutions are shown to depend continuously on the data of the problem; furthermore, they are well approximated by vanishing viscosity and regular approximations.
Probability,Analysis of PDEs
What problem does this paper attempt to address?
The problem that this paper attempts to solve is whether introducing rough Kraichnan noise into the generalized surface quasi - geostrophic equation (gSQG equation) can regularize these equations, thereby providing the existence and path - uniqueness of solutions under the initial data \(\theta_0\in L^1\cap L^p\). Specifically, the paper explores whether, by adding a specific type of stochastic transport noise (i.e., the Stratonovich transport noise of Kraichnan type), the generalized surface quasi - geostrophic equation can have the existence and path - uniqueness of strong solutions under certain parameter conditions. ### Background and Motivation The generalized surface quasi - geostrophic equation (gSQG equation) is a class of two - dimensional active scalar partial differential equations, in the form of: \[ \begin{cases} \partial_t\theta + u\cdot\nabla\theta = f,\\ u = -\nabla^{\perp}\Lambda^{- 2+\beta}\theta, \end{cases} \] where \(\Lambda = (-\Delta)^{1/2}\), \(\beta\in(0,1)\) is a parameter, interpolating between the 2D Euler equation (\(\beta = 0\)) and the SQG equation (\(\beta = 1\)). For the deterministic gSQG equation, it is known that there exist weak solutions \(\theta\in L^1\cap L^p\), but its uniqueness is an open problem. ### Research Objectives The main objective of the paper is to regularize the gSQG equation by introducing rough Kraichnan noise, specifically studying the following points: 1. **Existence of solutions**: Prove that after introducing noise, for the initial data \(\theta_0\in L^1\cap L^p\) and the external force \(f\in L^1_t(L^1\cap L^p)\), there exists a global probabilistic strong solution \(\theta\). 2. **Path - uniqueness**: Prove that under certain conditions, the path - uniqueness and law - uniqueness of solutions. 3. **Continuous dependence of solutions**: Prove the continuous dependence of solutions on initial values and external forces. 4. **Regularization effect**: Explore the regularization effect of noise on solutions, especially for different parameters \(\alpha\) and \(\beta\). ### Main Results The main results of the paper are as follows: - For the parameters \(\alpha,\beta,p\) satisfying the conditions \(0 <\frac{\beta}{2}<\alpha <\frac{1}{2}\) and \(\frac{\beta}{2}+\alpha\leq1-\frac{1}{p}\), \(p\in[2,\infty)\), for any \(\theta_0\in L^1\cap L^p\) and \(f\in L^1_{\text{loc}}([0,+\infty);L^1\cap L^p)\), there exists a global probabilistic strong solution \(\theta\), and satisfies \(\theta\in L^\infty(\Omega;L^\infty([0,T];L^1\cap L^p))\). - The path - uniqueness and law - uniqueness hold in the above solution class. - The continuous dependence of solutions on initial values and external forces is proved. Specifically, if \(\theta^n_0\rightarrow\theta_0\) in \(L^1\cap L^p\), \(\{f_n\}\) is bounded in \(L^1_{\text{loc}}([0,+\infty);L^1\cap L^p)\) and \(f_n\rightarrow f\) in \(L^1_{\text{loc}}([0,+\infty);\dot{H}^{\beta/2 - 1})\), then the corresponding solutions \(\theta_n\) converge in probability to \(\theta\).