Stochastic Lagrangian perturbation of Lie transport and applications to fluids (Q6155679)

The author considers a smooth and closed Riemannian manifold \(M\) of dimension \(n\), a smooth time-dependent family of exact exterior differential \(p\)-forms \(\beta _{t}=d\alpha _{t}\) on \(M\), where \(\alpha _{t}\) is a \((p-1)\) -form and \(d\) is the exterior derivative, and \(m+1\) smooth time-dependent vector fields \(\{V_{l}(t)\}_{l=0}^{m}\) on \(M\), satisfying \(V_{0}\in C([0,T]]; \mathfrak{X}^{k}(M))\) with components \(V_{0}^{i}\in L^{\infty }([0,T];C^{k,\gamma }(M))\), for \(i\in \{1,\ldots ,n\}\), \(V_{l}\in C([0,T]; \mathfrak{X}^{k+1}(M))\) with components \(V_{l}^{i}\in L^{\infty }([0,T];C^{k+1,\gamma }(M))\), for \(i\in \{1,\ldots ,n\}\) and \(l\in \{1,\ldots ,m\}\). Here \(\mathfrak{X}^{k}(M)\) is the space of \(C^{k}\)-vector field on \(M\). The linear differential operator \(L\) defined by \(L=\frac{1}{2} \sum_{l=1\mathfrak{L} V_{l}}^{m}\mathfrak{L}_{V_{l}}-\mathfrak{L} _{V_{0}}\) is supposed to be uniformly elliptic on \(M\), where \(\mathfrak{L} _{X}\) is the Lie derivative. The main result proves that if \(\xi _{s,t}(x)\) is the unique solution to the forward Stratonovich's stochastic differential equation: \( f(\xi _{s,t}(x))=f(x)+\int_{s}^{t}F(\circ dr)f(\xi _{s,r}(x))=f(x)+\int_{s}^{t}drV_{0}(r)f(\xi _{s,r}(x))+\sum_{l=1}^{m}\int_{s}^{t}V_{l}(r)f(\xi _{s,r}(x))\circ dW_{l}(r)\) , and \(\xi _{t,s}=\xi _{s,t}^{-1}\) its inverse, then \(\beta \) satisfies the dissipative transport equation: \(\partial _{t}\beta _{t}(x)+\mathfrak{L} V_{0}(t,x)\beta _{t}(x)-\frac{1}{2}\sum_{l=1}^{m}\mathfrak{L} V_{l}(t,x) \mathfrak{L} V_{l}(t,x)\beta _{t}(x)=0\), \(x\in M\), \(t\in ]0,T]\), with \(\beta _{0}(x)=\beta _{in}(x)\) as initial data, if and only if, for all \(x\in M\), and for all \(s\) such that \(0\leq s\leq t\leq T\), the statistical generalized Cauchy invariants equation \(\beta _{t}(x)=\mathbb{E}[d(\alpha _{i_{1}}\ldots _{i_{p-1}}(s)\circ \xi _{t,s}(x))\wedge d\xi _{t,s}^{i_{1}}(x)\wedge \ldots \wedge d\xi _{t,s}^{i_{p-1}}(x)]\) is satisfied. For the proof, the author uses a Lagrangian formulation of the Lie transport: Let \(\xi_{ s,t}(x)\) be the solution to the forward ordinary differential equation: \(\frac{d}{dt}\xi _{s,t}(x)=V_{0}(t,\xi _{s,t}(x))\), \(0\leq s\leq t\leq T\), \(\xi _{s,s}(x)=x\in M\), and \(\beta =d\alpha \in L^{\infty }([0,T];C^{1,\gamma }\Lambda ^{p}(M))\cap Lip([0,T];C^{0,\gamma }\Lambda ^{p}(M))\) be a time-dependent family of exact \(p\)-forms, with \(\alpha \in L^{\infty }([0,T];C^{2,\gamma }\Lambda ^{p-1}(M))\cap Lip([0,T];C^{1,\gamma }\Lambda ^{p-1}(M))\), then \(\beta \) satisfies the Lie transport equation \(\partial _{t}\beta _{t}(x)+\mathfrak{L} V_{0}(t,x)\beta _{t}(x)=0\), \(x\in M\), \(t\in ]0,T]\) , with \(\beta _{0}(x)=\beta _{in}(x)\) as initial data, if and only if, for all \(x\in M\), and for all \(s\) such that \(0\leq s\leq t\leq T\), the following backward generalized Cauchy invariants equation: \(\beta _{t}(x)=d(\alpha _{i_{1}}\ldots _{i_{p-1}}(s)\circ \xi _{t,s}(x))\wedge d\xi _{t,s}^{i_{1}}(x)\wedge \ldots \wedge d\xi _{t,s}^{i_{p-1}}(x)\) is satisfied. From this second result, the author derives the Kelvin circulation theorem and the Helmholtz flux conservation theorem. From the main result, he derives a Lagrangian formulation of non-ideal hydrodynamic and magnetohydrodynamic models and an analysis of the stochastic-Lagrangian incompressible extended MHD equations. He illustrates his results with different examples.