Anomalous dissipation and lack of selection in the Obukhov-Corrsin theory of scalar turbulence (Q6089031)

This paper studies anomalous dissipation and non uniqueness phenomena in linear transport theory, for the equation \[ \partial_t \theta + u \cdot \nabla \theta = 0 \quad \text{ in } \quad ]0,T[ \times \mathbb{T}^2 \, , \tag{1} \] posed for a scalar variable \(\theta:]0,T[\times \mathbb{T}^2\rightarrow \mathbb{R}\) and the velocity field \(u:]0,T[\times \mathbb{T}^2\rightarrow \mathbb{R}^2\), where \(\mathbb{T}^2 \) denotes the two-dimensional torus and \((0,T)\) is the time-interval. Periodic spacial extension is assumed and \(\theta\) is additionally subject to the initial condition \(\theta = \theta_{\mathrm{in}}\) at initial time. The velocity field is assumed to be Hölderian in space and, according to the context, either \(p\)-integrable or Hölderian in time, this means \(u \in L^p(]0,T[ ; C^{\alpha}(\mathbb{T}^2))\) or \(u \in C^{\alpha}(]0,T[\times \mathbb{T}^2)\). Moreover, it ought to be divergence-free in the weak sense. An important concern of the paper is to study in connection with (1) the asymptotic behaviour of the vanishing diffusivity solutions \(\theta = \theta_{\kappa}\) to the parabolic equation \[ \partial_t \theta + u \cdot \nabla \theta = \kappa \, \Delta \theta \quad \text{ in } \quad ]0,T[ \times \mathbb{T}^2 \, , \tag{2} \] subject to periodic boundary conditions and the initial condition \(\theta_{\mathrm{in}}\). In a first Theorem, the authors show that for \(p \geq 2\) and \(0 \leq \alpha <1\), there exist a velocity field \(u \in L^p(]0,T[ ; C^{\alpha}(\mathbb{T}^2))\) and an initial value \(\theta_{\mathrm{in}} \in C^{\infty}(\mathbb{T}^2)\) such that with \(p^{\circ} = 2p/(p-1)\) and any \(0 \leq \beta < 1/2\) satisfying \(\alpha + 2\beta < 1\), the problem (1) admits a solution \(\theta \in L^{p^{\circ}}(0,T; \, C^{\beta}(\mathbb{T}^2))\) that strictly dissipates the \(L^2\)-norm, in the sense that \(\|\theta(T,\cdot)\|_{L^2(\mathbb{T}^2)} < \|\theta_{\mathrm{in}}\|_{L^2(\mathbb{T}^2)}\). This solution can be gained as the limit of the sequence \(\{\theta_{\kappa}\}\) of solutions to (2) that exhibit the phenomenon anomalous dissipation: \[ \limsup_{\kappa \rightarrow 0} \kappa \, \int_0^T\int_{\mathbb{T}^2} |\nabla \theta_{\kappa}|^2 \, dxdt > 0 \, . \] In a second surprising Theorem B, the authors moreover show that for every \(\alpha \in [0,1[\) there are a velocity field \(u \in C^{\alpha}(]0,T[ \times \mathbb{T}^2)\) and initial data \(\theta_{\mathrm{in}} \in C^{\infty}(\mathbb{T}^2)\) such that the problem (1) possesses at least two solutions, the one of which conserves the \(L^2\)-norm, while the other will not. Both solutions are accumulation points of the sequence \(\{\theta_{\kappa}\}\) of solutions to (2), the first without and the second exhibiting anomalous dissipation. In this sense, the vanishing diffusion limit would not provide in its own a valid selection criterium for (1). The theorems rely on concrete, though highly technical, construction techniques for the singular velocity field and the initial data, while the convergence analysis is based on the introduction of an appropriate stochastic flow and the associated Feynman-Kac representation of the solution to (2).