The Dirichlet problem for the nonstationary Stokes system in a polygon
Jürgen Rossmann · 2025
Read The Dirichlet problem for the nonstationary Stokes system in a polygon by Jürgen Rossmann (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2503-01502.txt.
Introduction
Juergen Rossmann proves that the time-dependent Stokes system, the linearized equations of viscous flow, has strong solutions when the flow region is a polygon. The boundary values are prescribed. Corners are the enemy of regularity theory, since solutions can misbehave near sharp angles. Writing in twenty twenty-five, Rossmann works in weighted Sobolev spaces, where the weight is a power of the distance to the corner points. Allowed badness near corners is measured and controlled rather than ignored, and in that setting he establishes existence of strong solutions.
Catalog record
| CREATOR | Jürgen Rossmann |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 40 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2503-01502
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
The Dirichlet problem for the nonstationary Stokes system in a polygon Jürgen Rossmann Abstract 1 Introduction Let \Omega be a polygonal domain with the boundary \partial\Omega=\Gamma\cup{\cal P} , where {\cal P}=\{P_{1},\ldots,P_{n}\} is a set of corner points and \Gamma is the union of the smooth ( \in C^{2} ) arcs \Gamma_{1}=P_{1}P_{2},\ldots,\Gamma_{n-1}=P_{n-1}P_{n},\Gamma_{n}=P_{n}P_{1} . For simplicity, we assume that \Omega coincides with an angle K_{j} with opening \alpha_{j} in a neighborhood of the corner point P_{j} , j=1,\ldots,n . The present paper deals with the initial-boundary value problem \displaystyle\frac{\partial{\mathfrak{u}}}{\partial t}-\Delta{\mathfrak{u}}-\nabla\nabla\cdot{\mathfrak{u}}+\nabla{\mathfrak{p}}={\mathfrak{f}},\quad-\nabla\cdot{\mathfrak{u}}={\mathfrak{g}}\ \mbox{ in }Q_{T}=\Omega\times(0,T), (1) \displaystyle{\mathfrak{u}}(x,t)=0\ \mbox{ for }x\in\partial\Omega\backslash{\cal P},\ 0<t<T, (2) \displaystyle{\mathfrak{u}}(x,0)=0\ \mbox{ for }x\in\Omega. (3) The analogous problem in an infinite angle was already studied in [ 7 ] . As in [ 7 ] , the major part of the paper deals with the parameter-depending problem \displaystyle s\,u-\Delta u-\nabla\nabla\cdot u+\nabla p=f,\quad-\nabla\cdot u=g\mbox{ in }\Omega, (4) \displaystyle u=0\ \mbox{ on }\Gamma. (5) We are interested in solutions of this problem in the class of the weighted Sobolev spaces V_{\beta}^{l}(\Omega) which are defined for nonnegative integer l and real n -tuples \beta=(\beta_{1},\ldots,\beta_{n}) as the spaces (closure of the set C_{0}^{\infty}(\overline{\Omega}\backslash{\cal P}) with the norm \|u\|_{V_{\beta}^{l}(\Omega)}=\Big(\int_{\Omega}\sum_{|\alpha|\leq l}\prod_{j=1}^{n}r_{j}^{2(\beta_{j}-l+|\alpha|)}\,|\partial_{x}^{\alpha}u(x)|^{2}\,dx\Big)^{1/2}, where r_{j}(x) denotes the distance of the point x from the corner point P_{j} . We suppose that f\in V_{\beta}^{0}(\Omega) and g\in V_{\beta}^{1}(\Omega) , where the inequalities \max\big(1-\mbox{Re}\,\lambda^{*}(\alpha_{j}),-\frac{\pi}{\alpha_{j}}\big)<\beta_{j}<\min\big(1,\frac{\pi}{\alpha_{j}}\big),\quad\beta_{j}\not=0 (6) are satisfied for the components \beta_{j} of \beta . Here \lambda^{*}(\alpha) is the solution of the equation \sin(\lambda\alpha)+\lambda\sin\alpha=0 (7) with smallest positive real part. It is shown in Section 3 (see Theorem 3.1 ) that the problem ( 4 ), ( 5 ) has a unique solution (u,p) satisfying the condition \int_{\Omega}p(x)\,dx=0 and the estimate \|u\|_{V_{\beta}^{2}(\Omega)}+|s|\,\|u\|_{V_{\beta}^{0}(\Omega)}+\|p\|_{W_{\beta}^{1}(\Omega)}\leq c\,\Big(\|f\|_{V_{\beta}^{0}(\Omega)}+\|g\|_{V_{\beta}^{1}(\Omega)}+\ |s|\,\|g\|_{(W_{-\beta}^{1}(\Omega))^{*}}\Big) with a constant c independent of f,g and s provided that \mbox{Re}\,s\geq 0 and |s|>\gamma_{0} , where \gamma_{0} is a sufficiently large positive number. Here W_{\beta}^{1}(\Omega)=\{p\in L_{2}(\Omega):\ \nabla p\in V_{\beta}^{1}(\Omega)\} . Applying the inverse Laplace transform, we prove first the existence of solutions of the problem ( 1 )–( 3 ) for T=\infty in weighted spaces with the additional weight function e^{-\gamma t} , \gamma>\gamma_{0} (see Theorem 4.1 ). After this it is not difficult to prove the solvability of this problem on a finite t -interval. The main result of the paper (Theorem 4.2 ) ist the following: Suppose that {\mathfrak{f}}\in L_{2}(0,T;V_{\beta}^{0}(\Omega)) , {\mathfrak{g}}\in L_{2}(0,T;V_{\beta}^{1}(\Omega)) and \partial_{t}{\mathfrak{g}}\in L_{2}(0,T;(W_{-\beta}^{1}(\Omega))^{*}) , where the inequalities ( 6 ) are satisfied for all j . Furthermore, we assume that {\mathfrak{g}}(x,0)=0 for x\in\Omega and \int_{\Omega}{\mathfrak{g}}(x,t)\,dx=0\ \mbox{ for almost all }t. Then there exists a uniquely determined solution ({\mathfrak{u}},{\mathfrak{p}}) of the problem ( 1 )–( 3 ) such that {\mathfrak{u}}\in L_{2}(0,T;V_{\beta}^{2}(\Omega)) , {\mathfrak{u}}_{t}\in L_{2}(0,T;V_{\beta}^{0}(\Omega)) , {\mathfrak{p}}\in L_{2}(\Omega\times(0,T)) , \nabla{\mathfrak{p}}\in L_{2}(0,T;V_{\beta}^{0}(\Omega)) and \int_{\Omega}{\mathfrak{p}}(x,t)\,dx=0\ for\ almost\ all\ t. 2 The parameter-dependent problem in an angle Here, we present a theorem which was proved in [ 7 , Theorem 4.12 and Corollary 4.13] . Let K be an angle with aperture \alpha , i.e., K=\{x=(x_{1},x_{2}):\ 0<r<\infty,\ 0<\varphi<\alpha\} , where r , \varphi denote the polar coordinates of the point x . The sides of K are the half-lines \Gamma_{1} and \Gamma_{2} , where \varphi=0 on \Gamma_{1} and \varphi=\alpha on \Gamma_{2} . We consider the problem \displaystyle s\,u-\Delta u-\nabla\nabla\cdot u+\nabla p=f,\quad-\nabla\cdot u=g\mbox{ in }K, (8) \displaystyle u=0\mbox{ on }\Gamma_{1}\cup\Gamma_{2}. (9) For nonnegative integer l and real \beta , we define the weighted Sobolev spaces V_{\beta}^{l}(K) as the set of all functions (or vector functions) with finite norm \|u\|_{V_{\beta}^{l}(K)}=\Big(\int_{K}r^{2(\beta-l+|\alpha|)}\,\big|\partial_{x}^{\alpha}u(x)\big|^{2}\,dx\Big)^{1/2}. (10) The intersection V_{\beta}^{2}(K)\cap V_{\beta}^{0}(K) is denoted by E_{\beta}^{2}(K) . Furthermore, let W_{\beta}^{1}(K) be the space with the norm \|g\|_{W_{\beta}^{1}(K)}=\Big(\int\limits_{\begin{subarray}{c}K\\ 1<|x|<2\end{subarray}}|g|^{2}\,dx+\int_{K}r^{2\beta}|\nabla g|^{2}\,dx\Big)^{1/2}. Theorem 2.1 Suppose that s\not=0 , \mbox{Re}\,s\geq 0 and \max(-\frac{\pi}{\alpha}\,,1-\mbox{Re}\,\lambda^{*}(\alpha))<\beta<\min(2,\frac{\pi}{\alpha}) , \beta\not=0 , where \lambda^{*}(\alpha) is the solution of the equation ( 7 ) with smallest positive real part. Furthermore, let f\in V_{\beta}^{0}(K) and g\in V_{\beta}^{1}(K)\cap(W_{-\beta}^{1}(K))^{*} . In the case 0<\beta<2 , we assume that the integral of g over K is zero. Then the the problem s\,u-\Delta u-\nabla\nabla\cdot u+\nabla p=f,\quad-\nabla\cdot u=g\mbox{ in }K,\quad u=0\mbox{ on }\Gamma_{1}\cup\Gamma_{2} (11) has a unique solution (u,p)\in E_{\beta}^{2}(K)\times V_{\beta}^{1}(K) satisfying the estimate \|u\|_{V_{\beta}^{2}(K)}+|s|\,\|u\|_{V_{\beta}^{0}(K)}+\|p\|_{V_{\beta}^{1}(K)}\leq c\,\Big(\|f\|_{V_{\beta}^{0}(K)}+\|g\|_{V_{\beta}^{1}(K)}+|s|\,\|g\|_{(W_{-\beta}^{1}(K))^{*}}\Big) (12) with a constant c independent of f , g and s . Note that \mbox{Re}\,\lambda^{*}(\alpha)>1/2 for \alpha<2\pi and \mbox{Re}\,\lambda^{*}(\alpha)>1 for \alpha<\pi . Remark 2.1 In [ 7 , Theorem 4.12] , the assertion of Theorem 2.1 was obtained under the assumption that f\in V_{\beta}^{0}(K) and g\in V_{\beta}^{1}(K)\cap(V_{-\beta}^{1}(K))^{*} . But this assumption is the same as in Theorem 2.1 if the integral of g is zero for 0<\beta<2 . Indeed, if \beta<0 , then it follows from Hardy’s inequality that the spaces W_{-\beta}^{1}(K) and V_{-\beta}^{1}(K) coincide. Suppose that \beta>0 , g\in V_{\beta}^{1}(K)\cap(V_{-\beta}^{1}(K))^{*} and \int_{K}g\,dx=0 . Then any q\in W_{-\beta}^{1}(K) is continuous at the corner point x=0 and \displaystyle\Big|\int_{K}g\,q\,dx\Big| \displaystyle= \displaystyle\Big|\int_{K}g\,(q-q(0))\,dx\Big|\leq\|g\|_{(V_{-\beta}^{1}(K))^{*}}\ \|q-q(0)\|_{V_{-\beta}^{1}(K)} \displaystyle\leq \displaystyle c_{1}\,\|g\|_{(V_{-\beta}^{1}(K))^{*}}\ \|q\|_{W_{-\beta}^{1}(K)} (see [ 3 , Lemma 7.1.3] ). This means that g\in(W_{-\beta}^{1}(K))^{*} , \|g\|_{(W_{-\beta}^{1}(K))^{*}}\leq c_{1}\,\|g\|_{(V_{-\beta}^{1}(K))^{*}}\leq c_{2}\,\|g\|_{(W_{-\beta}^{1}(K))^{*}}\,, and the norm of g in (W_{-\beta}^{1}(K))^{*} in the estimate ( 12 ) can be replaced by the norm in the space (V_{-\beta}^{1}(K))^{*} . Remark 2.2 Suppose that f\in V_{\beta}^{0}(K)\cap V_{\gamma}^{0}(K) , g\in V_{\beta}^{1}(K)\cap V_{\gamma}^{1}(K) and g\in\cap(W_{-\beta}^{1}(K))^{*}\cap(W_{-\gamma}^{1}(K))^{*} , where \max\Big(-\frac{\pi}{\alpha}\,,1-\mbox{Re}\,\lambda^{*}(\alpha)\Big)<\beta<\gamma<\min\Big(2,\frac{\pi}{\alpha}\Big), \gamma>0 and \beta\not=0 . Furthermore, it is assumed that the integral of g over K is zero. Then there exist solutions (u,p)\in E_{\gamma}^{2}(K)\times V_{\gamma}^{1}(K) and (u^{\prime},p^{\prime})\in E_{\beta}^{2}(K)\times V_{\beta}^{1}(K) of the problem ( 11 ). By [ 7 , Lemma 4.5 and Lemma 4.7] , we have u=u^{\prime} and, consequently, \nabla p=\nabla p^{\prime} . If \beta>0 then even p=p^{\prime} . Thus, the solution (u,p) satisfies the estimate \|u\|_{V_{\beta}^{2}(K)}+|s|\,\|u\|_{V_{\beta}^{0}(K)}+\|\nabla p\|_{V_{\beta}^{0}(K)}\leq c\,\Big(\|f\|_{V_{\beta}^{0}(K)}+\|g\|_{V_{\beta}^{1}(K)}+|s|\,\|g\|_{(W_{-\beta}^{1}(K))^{*}}\Big) with a constant c independent of f , g and s . 3 The parameter-dependent problem in a polygon We consider the problem ( 4 ), ( 5 ) in the polygonal domain \Omega which was described in the introduction. First, we prove the existence of solutions (u,p)\in V_{\beta}^{2}(\Omega)\times W_{\beta}^{1}(\Omega) . After this, we estimate this solution in relation to the data f,g and the parameter s . 3.1 Existence of solutions Let W^{l}(\Omega) be the Sobolev space with the norm \|u\|_{W^{l}(\Omega)}=\Big(\int_{\Omega}\sum_{|\alpha|\leq l}|\partial_{x}^{\alpha}u(x)|^{2}\,dx\Big)^{1/2}. The closure of the set C_{0}^{\infty}(\Omega) in this space is denoted by \stackrel{{\scriptstyle\circ}}{{W}}\!{}^{l}(\Omega) , and W^{-l}(\Omega) is defined as the dual space of \stackrel{{\scriptstyle\circ}}{{W}}\!{}^{l}(\Omega) . By (\cdot,\cdot)_{\Omega} we denote the L_{2}(\Omega) -scalar product and its extension to the product W^{-l}(\Omega)\times\stackrel{{\scriptstyle\circ}}{{W}}\!{}^{l}(\Omega) . Furthermore, let \stackrel{{\scriptstyle\circ}}{{L}}{}_{2}(\Omega) be the set of all g\in L_{2}(\Omega) satisfying the condition \int_{\Omega}g(x)\,dx=0. (13) We consider the bilinear form b_{s}(u,v)=\int_{\Omega}\Big(su\cdot v+2\sum_{i,j=1}^{3}\varepsilon_{i,j}(u)\,\varepsilon_{i,j}(v)\Big)\,dx, where \varepsilon(u) denotes the strain tensor with the elements \varepsilon_{i,j}({\mathfrak{u}})=\frac{1}{2}\,\Big(\frac{\partial u_{i}}{\partial x_{j}}+\frac{\partial u_{j}}{\partial x_{i}}\Big),\quad i,j=1,\ldots,2, Obviously, this bilinear form is continuous on W^{1}(\Omega)\times W^{1}(\Omega) . Furthermore, it follows from Korn’s inequality that \big|b_{s}(u,\overline{u})\big|\geq c\,\|u\|^{2}_{W^{1}(\Omega)} (14) for all u\in W^{1}(\Omega) and for all s , \mbox{Re}\,s\geq 0 , s\not=0 , where c depends on |s| but not on u . The following lemma can be found e.g. in [ 2 , Chapter 1, Corollary 2.4] or [ 6 , Lemma 11.1.1] Lemma 3.1 Let g\in\stackrel{{\scriptstyle\circ}}{{L}}\!{}_{2}(\Omega) . Then there exists a vector function u\in\stackrel{{\scriptstyle\circ}}{{W}}\!{}^{1}(\Omega) such that \nabla\cdot u=g in \Omega and \|u\|_{W^{1}(\Omega)}\leq c\,\|g\|_{L_{2}(\Omega)} with a constant c independent of g . Applying the inequality ( 14 ) and Lemma 3.1 , one can prove the following lemma analogously to [ 6 , Theorem 11.1.2] . Lemma 3.2 Suppose that f\in W^{-1}(\Omega) , g\in\stackrel{{\scriptstyle\circ}}{{L}}\!{}_{2}(\Omega) , \mbox{Re}\,s\geq