Large deviation principles for 3D stochastic primitive equations (Q2358728)

From MaRDI portal
scientific article
Language Label Description Also known as
English
Large deviation principles for 3D stochastic primitive equations
scientific article

    Statements

    Large deviation principles for 3D stochastic primitive equations (English)
    0 references
    0 references
    0 references
    0 references
    15 June 2017
    0 references
    Let \(D\) be a bounded open domain in \(\mathbb R^2\) with smooth boundary \(\partial D\). The authors consider 3D primitive equations \[ \begin{aligned} \frac{\partial v}{\partial t}+(v\cdot\nabla)v+\theta\frac{\partial v}{\partial z}+fk\times v+\nabla P+Lv&=\psi_1(t,v,T)\frac{dW_1}{\,dt}, \\ \frac{\partial T}{\partial t}+(v\cdot\nabla)T+\theta\frac{\partial T}{\partial z}+LT&=\psi_2(t,v,T)\frac{dW_2}{dt}, \\ \partial_zP+T&=0, \\ \nabla\cdot v+\partial_z\theta&=0, \end{aligned} \] on \(\mathcal O=D\times(-1,0)\) subject to boundary conditions \[ \begin{aligned} \partial_zv=0, \;\partial_z T=0\quad&\text{on}\quad D\times\{0,1\}, \\ v=0, \;\frac{\partial T}{\partial n}=0\quad&\text{on}\quad\partial D\times[0,1]. \end{aligned} \] Here, \(v\) is the horizontal velocity field, \(\theta\) the vertical velocity field, \(f\) the Coriolis parameter, \(k\) the vertical unit vector, \(P\) the pressure, \(T\) the temperature, \(Lv=-\Delta v-\frac{\partial^2v}{\partial z^2}\), \(\psi\) is a diffusion nonlinearity, \(n\) the outward normal vector on \(\partial\mathcal O\) and \(W_1\), \(W_2\) are independent cylindrical Wiener processes on some Hilbert spaces \(U_1\) and \(U_2\), respectively. The quantities \(v\), \(\theta\), \(T\) and \(P\) are unknown. Let \(V\) and \(H\) be the closures of smooth functions subject to the boundary conditions for \((v,T)\) in the \(H^1\)-norm and \(L^2\)-norm, respectively, and \(A\) the linear operator on \(L^2\) corresponding to the bilinear form \[ ((v,\tilde v),(T,\tilde T))\mapsto\int_{\mathcal O}\left(\nabla_{(x,y,z)}v\cdot\nabla_{(x,y,z)}\tilde v+\nabla_{(x,y,z)}T\cdot\nabla_{(x,y,z)}\tilde T\right)\,dxdydz \] on \(V\times V\). Let \(Y_0\in V\) be fixed and, for \(\tilde h=\int_0^\cdot h\,ds\) for some \(h\in L^2([0,\tau];U_1\times U_2)\), let \(\mathcal G^0(\tilde h)\) denote the solution to the equation above with the initial condition \(Y_0\) and with the noise \(dW\) replaced by \(d\tilde h\). Set \(\mathcal G^0(\tilde h)=0\) for other \(\tilde h\). The main result of the paper states that solutions \(\{Y^\varepsilon\}\) to the equation with the initial condition \(Y_0\) and with the diffusion nonlinearity \(\psi\) replaced by \(\sqrt\varepsilon\psi\) satisfy the large deviation principle on \(C([0,\tau];V)\cap L^2([0,\tau];D(A))\) with the good rate function \[ I(w)=\inf\,\left\{\frac 12\int_0^\tau|h(s)|^2_{U_1\times U_2}\,ds:\,h\in L^2(0,\tau;U_1\times U_2),\,w=\mathcal G^0\left(\int_0^\cdot h(s)\,ds\right)\right\}. \]
    0 references
    stochastic primitive equations
    0 references
    large deviations
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references

    Identifiers