Motion in a Gaussian incompressible flow (Q1355745): Difference between revisions
From MaRDI portal
Removed claims |
Changed an Item |
||
Property / author | |||
Property / author: George S. Papanicolaou / rank | |||
Normal rank | |||
Property / reviewed by | |||
Property / reviewed by: Isamu Dôku / rank | |||
Normal rank |
Revision as of 12:06, 10 February 2024
scientific article
Language | Label | Description | Also known as |
---|---|---|---|
English | Motion in a Gaussian incompressible flow |
scientific article |
Statements
Motion in a Gaussian incompressible flow (English)
0 references
7 January 1998
0 references
Let us consider a particle undergoing diffusion with a drift caused by an external velocity field \({\mathbf V}(t,x)\), \(t>0\), \(x\in\mathbb{R}^d\). Its motion is described by the Itô SDE \[ d{\mathbf X}(t)= {\mathbf V}(t,{\mathbf X}(t))dt+ \sigma\cdot d{\mathbf B}(t),\quad {\mathbf X}(0)=\text{\textbf{0}},\tag{1} \] with a \(d\)-dimensional Brownian motion \({\mathbf B}(t)\), where \(\sigma^2\) is the molecular diffusivity of the medium. We assume that \({\mathbf V}\) is random, zero mean, divergence free and mixing at macroscopically short space and time scales, so that it may model turbulent, incompressible flows. To describe the long time and long distance behavior of the particle trajectory, we introduce macroscopic units in which the time \(t\) (resp. space \({\mathbf x}\)) variable is of order \(\varepsilon^{-2}\) (resp. \(\varepsilon^{-1}\)) with a small scaling parameter \(\varepsilon>0\). Then our main concern will be the asymptotic behavior of the trajectories \(\{{\mathbf X}_\varepsilon(t)\}_{t\geq 0}\) when \(\varepsilon\downarrow 0\). According to the mathematical theory of turbulent diffusion [e.g. \textit{K. Oelschläger}, Ann. Probab. 16, No. 3, 1084-1126 (1988; Zbl 0653.60047) and \textit{S. Molchanov}, in: Lectures on probability theory. École d'Été de Probabilités de Saint-Flour XXII - 1992, Lect. Notes Math. 1581, 242-411 (1994; Zbl 0814.60093)], it is shown that for an ergodic zero mean velocity field \({\mathbf V}\) with sufficiently smooth realizations and suitably restricted power spectra, when \(\sigma\) is positive, then \(\{{\mathbf X}_\varepsilon(t)\}\) converges weakly to a Brownian motion in dimensions \(d>2\). Its covariance matrix \({\mathbf K}= [\kappa_{i,j}]_{i,j=1,\dots,d}\) is given by \[ \kappa_{i,j}= \int^{+\infty}_0{\mathbf M}{\mathbf E}\{V_i(t,{\mathbf X}(t))V_j(0,\text{\textbf{0}})+ V_j(t,{\mathbf X}(t))V_i(0, \text{\textbf{0}})\}dt+ \sigma^2\delta_{i,j},\tag{2} \] where \({\mathbf X}(t)\) is the solution of (1) and \({\mathbf M}\) (resp. \({\mathbf E}\)) denotes the expectation over the underlying probability space for \({\mathbf B}(t)\) (resp. for \({\mathbf V}\)). Since the formula (2) for the effective diffusivity \({\mathbf K}\) involves the solution of (1), it is far from being explicit. In many applications it is important to consider how \({\mathbf K}\) behaves as a function of \(\sigma\), especially when \(\sigma\downarrow 0\). The authors go one step further and ask about the limiting behavior of the trajectories \({\mathbf X}_\varepsilon\) of SDE (1) with the scaled variables, as \(\varepsilon\downarrow 0\), in the absence of any molecular diffusivity, that is, when \(\sigma=0\). The effective diffusivity, if it exists in this case, is purely due to turbulent advection. It is this aspect of the problem of turbulent diffusion that the authors will address in this paper. On this account, we consider an ordinary differential equation \[ d{\mathbf X}_\varepsilon(t)/dt= \varepsilon^{-1}{\mathbf V}(t/\varepsilon^2,{\mathbf X}_\varepsilon(t)/\varepsilon),\quad {\mathbf X}_\varepsilon(0)= \text{\textbf{0}},\tag{3} \] where \({\mathbf V}\) is a \(d\)-dimensional, random, incompressible, stationary Gaussian field which has mean zero and decorrelates in finite time. With some additional conditions on \({\mathbf V}\), the authors prove that the family of trajectories \(\{{\mathbf X}_\varepsilon(t)\}\) converges weakly to a Brownian motion with covariance matrix given by the Kubo formula \[ \kappa_{i,j}= \int^\infty_0{\mathbf E}[V_i(t,\text{\textbf{0}})V_j(0,\text{\textbf{0}})+ V_j(t,\text{\textbf{0}}) V_i(0,\text{\textbf{0}})]dt, \] when \(\varepsilon\downarrow 0\). For other related results, see for instance \textit{R. Carmona} and \textit{L. Xu} [Ann. Appl. Probab. 7, No. 1, 265-279 (1997; Zbl 0879.60063)].
0 references
random field
0 references
mixing condition
0 references
weak convergence
0 references
diffusion approximation
0 references
asymptotic behavior of the trajectories
0 references
absence of any molecular diffusivity
0 references
Brownian motion
0 references
Kubo formula
0 references