Global solutions of the gravity-capillary water-wave system in three dimensions (Q1750668)

The authors consider the free-boundary incompressible Euler equations written as \((\partial _{t}+v\cdot \nabla )v=-\nabla p-ge_{n}\), \(\nabla \cdot v=0\), posed in a domain \(\Omega _{t}\subset \mathbb{R}^{n}\), \(n\geq 2\). In the case of an irrotational fluid \(\mathrm{curl } v=0\) and assuming that the domain \(\Omega _{t}\) may be written as \(\Omega _{t}=\{(x,y)\in \mathbb{R} ^{n-1}\times \mathbb{R}:y\leq h(x,t)\}\), with upper boundary \( S_{t}=\{(x,y):y=h(x,t)\}\), the authors observe that this problem may be transferred to a boundary one. They indeed introduce the velocity potential \(\Phi \) defined through \(\nabla _{x,y}\Phi (x,y,t)=v(x,y,)\) and its restriction to \(S_{t}\): \(\phi (x,y,t)=\Phi (x,h(x,t),t)\). They then observe that the above problem may be written as \(\partial _{t}h=G(h)\phi \), \( \partial _{t}\phi =-gh+\sigma \operatorname{div} (\frac{\nabla h}{\sqrt{1+\left| \nabla h\right| ^{2}}})-\frac{1}{2}\left| \nabla \phi \right| ^{2}+\frac{(G(h)\phi +\nabla h\cdot \nabla \phi )^{2}}{2(1+\left| \nabla h\right| ^{2})}\), where \(G(h)=\sqrt{1+\left| \nabla h\right| ^{2} }\mathcal{N}(h)\), \(\mathcal{N}(h)\) being the Dirichlet-Neumann map associated with the domain \(\Omega _{t}\), \(g\) is the gravitational constant and \(\sigma \geq 0\) is the surface tension coefficient. Initial data \((h_{0},\phi _{0})\) are imposed. The authors associate to this last problem the energy functional \(\mathcal{H}(h,\phi )=\frac{1}{2}\int_{\mathbb{R}^{n-1}}G(h)\phi \cdot \phi dx+\frac{g}{2}\int_{\mathbb{R}^{n-1}}h^{2}dx+\sigma \int_{\mathbb{ R}^{n-1}}\frac{\nabla h}{1+\sqrt{1+\left| \nabla h\right| ^{2}}}dx\). The main result of the paper presents a global regularity result in the case of the gravity-capillary water-wave system obtained when \(g>0\) and \(\sigma >0 \) and in the 3D case. Assuming that the initial data \(\mathcal{U} _{0}=(g-\sigma \Delta )^{1/2}h_{0}+i\left| \nabla \right| ^{1/2}\phi _{0}\) satisfies \(\left\| \mathcal{U}_{0}\right\| _{H^{N_{0}}\cap H_{\Omega }^{N_{1},N_{3}}}+\sup_{2m+\left| \alpha \right| \leq N_{1}+N_{4}}\left\| (1+\left| x\right| )^{1-50\delta }D^{\alpha }D^{m}\mathcal{U}_{0}\right\| _{L^{2}}=\varepsilon _{0}\leq \overline{ \varepsilon }_{0}\), for \(\delta \) and \(\overline{\varepsilon }_{0}\) small enough and \(N_{0}\), \(N_{1}\), \(N_{3}\), and \(N_{4}\) large enough, they prove that the preceding problem has a unique solution \((h,\phi )\in C([0,\infty ):H^{N_{0}+1}\times \dot H^{N_{0}+1/2,1/2}\) which satisfies \( (1+t)^{-\delta ^{2}}\left\| \mathcal{U}(t)\right\| _{H^{N_{0}}\cap H_{\Omega }^{N_{1},N_{3}}}\precsim \varepsilon _{0}\) and \((1+t)^{5/6-3\delta ^{2}}\left\| \mathcal{U}(t)\right\| _{L^{\infty }}\precsim \varepsilon _{0}\), where \(\mathcal{U}=(g-\sigma \Delta )^{1/2}h+i\left| \nabla \right| ^{1/2}\phi \). Here the spaces are equipped with the norms \( \left\| f\right\| _{H_{\Omega }^{N^{\prime },N}}=\sum_{j\leq N^{\prime }}\left\| \Omega ^{j}f\right\| _{H^{N}}\) with \(\Omega =x_{1}\partial _{x_{2}}-x_{2}\partial _{x_{1}}\) and \(\left\| f\right\| _{\dot H^{N,b}}=\left\| (\left| \nabla \right| ^{N}+\left| \nabla \right| ^{b})f\right\| _{L^{2}}\). The authors first establish a local in time existence result. Assuming that \(g=1=\sigma \) and \(\left\| h_{0}\right\| _{H^{N+1}}+\left\| \phi _{0}\right\| _{\dot H ^{N+1/2,1.2}}\) is small enough, with \(N\geq 10\), they prove the existence of a solution to the preceding system \((h,\phi )\in C([0,1]:H^{N+1}\times\dot H^{N+1/2,1/2}\). They also prove that the profile \(\mathcal{V} (t):=e^{it\Lambda }\mathcal{U}(t)\), with \(\Lambda (\xi )=\sqrt{\left| \xi \right| +\left| \xi \right| ^{3}}\), of the solution \( \mathcal{U}(t)\) is continuous on an interval \([0,T_{0}]\setminus [ t_{0},t_{0}+\tau ]\) for \(T_{0}\geq 1\) and some \(\tau \) with respect to an appropriate norm. Working in the Fourier space, they then improve energy estimates on the function \(\mathcal{U}\), using the the semilinear structure of space-time integrals and dissipative estimates on the profile \(\mathcal{V} \), using bootstrap arguments. They also analyze a non-degeneracy property on the function \(\Upsilon (\xi ,\eta )=\nabla _{\xi ,\eta }^{2}\Phi (\xi ,\eta )[\nabla _{\xi }^{\perp }\Phi (\xi ,\eta )+\nabla _{\eta }^{\perp }\Phi (\xi ,\eta )]\). The proofs are given with details, as the authors propose new tools to study such problems and the long paper ends with further complements on the paradifferential calculus and the properties of the Dirichlet-Neumann operator.