A stochastically perturbed fluid-structure interaction problem modeled by a stochastic viscous wave equation (Q2064303)

The authors derive a stochastic model consisting of a stochastic viscous wave equation written as: \(u_{tt}+2\mu \sqrt{-\Delta}u_{t}-\Delta u=f(u)W(dt,dx)\), and posed in \(\mathbb{R}^{n}\), \(n=1,2\). Here \(W(dt,dx)\) is a noise which accounts for stochastic effects and \(f(u)\) is a nonlinear and Lipschitz continuous function. The initial conditions \(u(0,x)=g(x)\) and \(\partial_{t}u(0,x)=h(x)\) are imposed, \(g\) and \(h\) being continuous versions of functions in \(H^{2}(\mathbb{R}^{n})\). The authors define a mild solution to this problem as a stochastic process \(u(t,x)\) which is jointly measurable and adapted to the filtration \(\mathcal{F}_{t}\) such that \[ u(t,x,\omega)=\int_{\mathbb{R}^{n}}J_{t}(x-y)g(y)dy+\int_{\mathbb{R}^{n}}K_{t}(x-y)h(y)dy+\int_{0}^{t}\int_{\mathbb{R}^{n}}K_{t-s}(x-y)f(u(s,y, \omega))W(dy,ds), \] where \[ J_{t}(x)=\frac{1}{(2\pi)^{n}}\int_{\mathbb{R}^{n}}e^{ix\cdot \xi}e^{-\frac{\left\vert \xi \right\vert}{2}t} \cos(\frac{ \sqrt{3}}{2}\left\vert \xi \right\vert t)+\frac{1}{\sqrt{3}} \sin(\frac{\sqrt{3}}{2}\left\vert \xi \right\vert t)d\xi \] and \[ K_{t}(x)=\frac{1}{(2\pi)^{n}} \int_{\mathbb{R}^{n}}e^{ix\cdot \xi}e^{-\frac{\left\vert \xi \right\vert}{2}t}\frac{\sin (\frac{\sqrt{3}}{2}\left\vert \xi \right\vert t)}{\frac{\sqrt{3}}{2}\left\vert \xi \right\vert}d\xi. \] The first main result of the paper proves the existence of a mild solution to this problem, which is unique up to stochastic modification. For the proof, the authors use Picard iterations, setting the first iterate \(u_{0}\) as: \[ u_{0}(t,x)=\int_{\mathbb{R}^{n}}J_{t}(x-y)g(y)dy+\int_{\mathbb{R}^{n}}K_{t}(x-y)h(y)dy, \] then defining \(u_{k}\) for \(k\geq 1\) as: \[ u_{k}(t,x)=u_{0}(t,x)+\int_{0}^{t}\int_{\mathbb{R}^{n}}K_{t-s}(x-y)f(u_{k-1}(s,y))W(dy,ds). \] They prove that this Picard iteration procedure is well defined at each step and they estimate the difference between consecutive iterates using properties of the kernels. They prove the convergence of the Picard iteration in \(L^{2}(\Omega)\) to some \(u(t,x)\) for each \(t\geq 0\) and \(x\in \mathbb{R}^{n}\). They finally prove that this limit \(u\) satisfies the preceding equation. For the uniqueness, the authors apply Gronwall's inequality. In the last part of their paper, the authors prove different Hölder and Kolmogorov continuity results. for this mild solution They especially prove that, up to a modification, the stochastic mild solution is \(\alpha \)-Hölder continuous for almost every realization of the solution sample path, where \(\alpha \in [0,1)\) if \(n=1\), and \(\alpha \in [0,1/2)\) if \(n=2\).