Weak solutions to incompressible heat-conducting motions with large flux
Joanna Rencławowicz, Wojciech M. Zaja̧czkowski · 2025
Read Weak solutions to incompressible heat-conducting motions with large flux by Joanna Rencławowicz, Wojciech M. Zaja̧czkowski (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-2506-22155.txt.
Introduction
Joanna Rencławowicz and Wojciech Zajączkowski prove existence results for a heat-conducting fluid in a 2025 paper. Their model joins the Navier-Stokes equations for viscous flow with the heat equation, set in a finite cylinder with strong inflow and outflow through its ends. Large fluxes make the mathematics hard, since boundary terms can overwhelm standard energy bounds.
The authors first prove an energy estimate that needs no cap on the size of the forcing, the starting data, or the inflow and outflow. Weighted function spaces give the grip needed on nonlinear terms. A Galerkin approximation then builds solutions step by step, and the estimate carries through the limit to show weak solutions exist. The result covers a regime earlier theory excluded.
Catalog record
| CREATOR | Joanna Rencławowicz, Wojciech M. Zaja̧czkowski |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 73 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2506-22155
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Weak solutions to incompressible heat-conducting motions with large flux Joanna Rencławowicz, Wojciech M. Zaja̧czkowski Abstract We analyze the problem for viscous incompressible heat-conducting fluid in a finite cylinder with large inflow and outflow, modelled with Navier-Stokes equations coupled with the heat equation. We prove energy estimate without restrictions on the magnitudes of the external force, initial data, inflow and outflow. In order to estimate nonlinear terms we use weighted spaces. Next, using Galerkin approximation, we show existence of weak solutions. 1) Institute of Mathematics, Polish Academy of Sciences, Śniadeckich 8, 00-656 Warsaw, Poland e-mail: [email protected], ORCID: 0000-0001-8597-9366, corresponding author, 2) Institute of Mathematics and Cryptology, Cybernetics Faculty, Military University of Technology, S. Kaliskiego 2, 00-908 Warsaw, Poland e-mail: [email protected], ORCID: 0000-0003-1229-2162, emeritus professor at IMPAN † † Keywords : Navier-Stokes equation, heat-conducting, weak solutions, energy estimate Mathematics Subject Classification (MSC2020) : Primary 35Q30; Secondary 76D03, 76D05 Abbreviated title: Weak solutions for heat-conducting flow. Statement relating to the ethics and integrity policies: On behalf of all authors, the corresponding author states that the manuscript is not published elsewhere and there is no conflict of interest. The datasets generated during and/or analysed during the current study are available from the corresponding author on reasonable request. Funding statement: There are no funders to report for this submission. 1 Introduction We consider the following motions of viscous incompressible heat-conducting fluid in a finite cylinder \Omega\subset{\mathbb{R}}^{3} with boundary S=S_{1}\cup S_{2} with arbitrary large inflow and outflow (1.1) \displaystyle{{}v_{t}+v\cdot\nabla v-{\rm div}\,{\mathbb{T}}(v,p)=\omega(\theta)f\ \ {\rm in}\ \ \Omega\times{\mathbb{R}}_{+},} \displaystyle{{}{\rm div}\,v=0\qquad{\rm in}\ \ \Omega\times{\mathbb{R}}_{+},} \displaystyle{{}\theta_{t}+v\cdot\nabla\theta-\mathchar 1403\relax\Delta\theta=0\qquad{\rm in}\ \ \Omega\times{\mathbb{R}}_{+},} \displaystyle{{}\nu\bar{n}\cdot{\mathbb{D}}(v)\cdot\bar{\tau}_{\alpha}+\gamma v\cdot\bar{\tau}_{\alpha}=0,\ \ \alpha=1,2,\qquad{\rm on}\ \ S_{1}\times{\mathbb{R}}_{+},} \displaystyle{{}v\cdot\bar{n}=0\qquad{\rm on}\ \ S_{1}\times{\mathbb{R}}_{+},} \displaystyle{{}\bar{n}\cdot{\mathbb{D}}(v)\cdot\bar{\tau}_{\alpha}=0,\ \ \alpha=1,2,\qquad{\rm on}\ \ S_{2}\times{\mathbb{R}}_{+},} \displaystyle{{}v\cdot\bar{n}=d\qquad{\rm on}\ \ S_{2}\times{\mathbb{R}}_{+},} \displaystyle{{}\bar{n}\cdot\nabla\theta=0\qquad{\rm on}\ \ S_{1}\times{\mathbb{R}}_{+},} \displaystyle{{}\bar{n}\cdot\nabla\theta=0\qquad{\rm on}\ \ S_{2}\times{\mathbb{R}}_{+},} \displaystyle{{}v|_{t=0}=v_{0},\ \ \theta|_{t=0}=\theta_{0}\qquad{\rm in}\ \ \Omega,} where \Omega is parallel to the x_{3} -axis of the Cartesian coordinates x=(x_{1},x_{2},x_{3}) . Moreover, S_{1} is parallel to the x_{3} -axis and S_{2} is perpendicular to it, v=(v_{1}(x,t),v_{2}(x,t),v_{3}(x,t))\in{\mathbb{R}}^{3} is the velocity of the fluid, p=p(x,t)\in{\mathbb{R}} is the pressure, \theta=\theta(x,t)\in{\mathbb{R}}_{+} is the temperature of the fluid, f=(f_{1}(x,t),f_{2}(x,t),f_{3}(x,t))\in{\mathbb{R}}^{3} is the external force field, \omega\in{\mathbb{R}}^{1} is the function of temperature, \bar{n} is the unit outward vector normal to S and \bar{\tau}_{\alpha} , \alpha=1,2 , are tangent vectors to S . By {\mathbb{T}}(v,p)=\nu{\mathbb{D}}(v)-p{\mathbb{I}} we denote the stress tensor, where \nu>0 is the constant viscosity coefficient, {\mathbb{I}} is the unit matrix, and {\mathbb{D}}(v)=\{v_{i,x_{j}}+v_{j,x_{i}}\}_{i,j=1,2,3} is the dilatation tensor. Additionally, \mathchar 1403\relax>0 is the constant heat conductivity coefficient, \gamma>0 is the slip coefficient. Let a , c_{0} be positive numbers and let \phi_{0}(x_{1},x_{2})=c_{0} be a sufficiently smooth closed curve (not a circle) in the plane x_{3}={\rm const}\in[-a,a] . Then \displaystyle{{}S_{1}=\{x\in{\mathbb{R}}^{3}\colon\phi_{0}(x_{1},x_{2})=c_{0},-a<x_{3}<a\},} \displaystyle{{}S_{2}=S_{2}(-a)\cup S_{2}(a),} \displaystyle{{}S_{2}(-a)=\{x\in{\mathbb{R}}^{3}\colon\phi_{0}(x_{1},x_{2})<c_{0},x_{3}=-a\},} \displaystyle{{}S_{2}(a)=\{x\in{\mathbb{R}}^{3}\colon\phi_{0}(x_{1},x_{2})<c_{0},x_{3}=a\}.} To describe inflow and outflow we denote d=(d_{1},d_{2}), d|_{S_{2}(a_{1})}=-d_{1},\quad d|_{S_{2}(a_{2})}=d_{2}, where d_{i}\geq 0 , i=1,2 , a_{1}=-a , a_{2}=a . The continuity equation (\ref{1.1})_{2} implies the compatibility condition (1.2) \intop_{S_{2}(-a)}d_{1}dS_{2}=\intop_{S_{2}(a)}d_{2}dS_{2}. In this paper we show the energy estimate to problem ( 1.1 ) without restrictions on the magnitudes of the external force f, initial data v(0),\theta(0), inflow d_{1} and outflow d_{2}. We assume that the flux does not have to vanish as t\rightarrow\infty. The result requires applying the Hopf function (see [ L1 ] ) and weighted estimates proved in [ RZ2 , RZ3 , RZ4 ] and recalled in Lemma 2.2 . Namely, we need to estimate nonlinear term caused by v\cdot\nabla v multiplied by test function in this way to get some smallness and to absorb this small term with norms on the l.h.s. Since the norm in L_{p} is not small, we use weighted norm, see Remark 3.3 for details. Now, we formulate the main result, established in Theorems 3.6 and 3.7 . Main Theorem . (Weak solutions and energy estimate) Let (v,p,\theta) be a solution to problem ( 1.1 ). Assume that d_{i}\in L_{\infty}(0,T;W_{3}^{1}(S_{2}(a_{i})))\cap L_{2}(0,T;H^{1/2}(S_{2}(a_{i}))),i=1,2, \theta(0)\in L_{2}(\Omega), \omega f\in L_{6/5}(\Omega^{T}). Then there exist weak solution (w,\theta) , where w is defined by ( 3.9 ), \vartheta is defined by ( 3.10 ) and (1.3) \displaystyle\begin{aligned} &\|v\|_{V(\Omega^{T})}^{2}+\|\theta\|_{V(\Omega^{T})}^{2}\leq A_{1}^{2}+\|w(0)\|_{L_{2}(\Omega)}^{2}+\|\vartheta(0)\|_{L_{2}(\Omega^{t})}^{2}\\ &+c\sup_{t^{\prime}\in(0,T)}\|\tilde{d}(t^{\prime})\|_{L_{2}(\Omega)}^{2}+c\intop_{0}^{T}\|\tilde{d}(t^{\prime})\|_{W^{1}_{2}(\Omega)}^{2}dt^{\prime}\\ &+c\sup_{t^{\prime}\in(0,T]}\|\tilde{d}(t^{\prime})\|_{W_{3,\infty}^{1}(\Omega)}^{4}\exp\bigg(\frac{1}{c}\sup_{t^{\prime}\in(0,T)}\|\tilde{d}(t^{\prime})\|_{W_{3,\infty}^{1}(\Omega)}\bigg)\cdot\\ &\quad\cdot\intop_{0}^{T}\sum_{i=1}^{2}\sup_{x_{3}}\|\tilde{d}_{i}(t^{\prime})\|_{L_{2}(S_{2}(a_{i}))}^{2}dt^{\prime}+c\|\omega(\theta)f\|_{L_{6/5}(\Omega)}^{2},\end{aligned} where \tilde{d}_{i} is an extension of d_{i} on \Omega , \tilde{d}_{i}|_{S_{2}(a_{i})}=d_{i}, i=1,2, d denotes (d_{1},d_{2}), \displaystyle||v||_{V(\Omega^{T})}\equiv{\rm ess}\sup_{t\in(0,T)}||v||_{L_{2}(\Omega)}+\left(\int_{0}^{T}||\nabla v||_{L_{2}(\Omega)}^{2}dt\right)^{1/2} and \displaystyle A_{1}^{2}=\int_{0}^{T}(\|\omega(\theta)f(t)\|_{L_{6/5}(\Omega)}^{2}+\sum_{i=1}^{2}\|d_{i}(t)\|_{W^{1}_{3}(S_{2}(a_{i}))}^{2}+\sum_{i=1}^{2}\|d_{i,t}(t)\|_{W^{1}_{6/5}(S_{2}(a_{i}))}^{2})dt\cdot \displaystyle\quad\,\,\cdot\phi(\sum_{i=1}^{2}\sup_{t}\|d_{i}(t)\|_{W^{1}_{3}(S_{2}(a_{i}))})<\infty, with an increasing positive function \phi. Moreover, \theta is bounded from above and below with positive \theta^{*} and \theta_{*}, respectively, \theta_{*}<\theta^{*} (1.4) \displaystyle\begin{aligned} &\|\theta\|_{V(\Omega^{T})}^{2}\leq c\exp(\|d_{1}\|_{L_{6}(0,T;L_{3}(S_{2}(a_{1})))}^{6})\|\theta(0)\|_{L_{2}(\Omega)}^{2}\\ &\quad+\|d_{1}\|^{2}_{L_{1}(S_{2}^{T}(a_{1}))}{\theta^{*}}^{2}+\|\theta(0)\|_{L_{2}(\Omega)}^{2}.\end{aligned} We show the existence of weak solutions to problem ( 1.1 ) through the Galerkin method, similarly as in [ ZZ ] . On the other hand, comparing results with [ ZZ ] , we can relax conditions on function \omega(\theta) since we prove additionally boundedness for temperature \theta from below an above, see Lemma 2.4 . 2 Notation and auxiliary results First we introduce a simplified notation of spaces used in this paper. Definition 2.1 (see [ Ad ] ) . For Lebesque and Sobolev spaces we set the following notation \displaystyle{{}\|u\|_{L_{p}(Q)}=|u|_{p,Q},\quad\|u\|_{L_{p}(Q^{t})}=|u|_{p,Q^{t}},} \displaystyle{{}\|u\|_{L_{q}(0,t;L_{p}(Q))}=|u|_{p,q,Q^{t}},} where Q is equal either \Omega , or S=\partial\Omega , or {\mathbb{R}}^{3} , Q^{t}=Q\times(0,t) , p,q\in[1,\infty] . Let H^{s}(Q)=W_{2}^{s}(Q) , s\in{\mathbb{N}}_{0}={\mathbb{N}}\cup\{0\} . Then we denote \displaystyle{{}\|u\|_{H^{s}(Q)}=\|u\|_{s,Q},\quad} \displaystyle{{}\|u\|_{W_{p}^{s}(Q)}=\|u\|_{s,p,Q},} \displaystyle{{}\|u\|_{L_{q}(0,t;W_{p}^{s}(Q))}=\|u\|_{s,p,q,Q^{t}},\quad} \displaystyle{{}\|u\|_{L_{q}(0,t;W_{q}^{s}(Q))}=\|u\|_{s,q,Q^{t}}.} Introduce weighted spaces L_{p,\mu}(\Omega),L^{k}_{p,\mu}(\Omega) , p\in[1,\infty] , \mu\in(0,1) , with the norms \|u\|_{L_{p,\mu}(\Omega)}=\bigg(\intop_{\Omega}|u|^{p}\eta^{p\mu}dx\bigg)^{1/p}, \displaystyle\|u\|_{L_{p,\mu}^{k}(\Omega)}=\bigg(\sum_{|\alpha|=k}\int_{\Omega}|D_{x}^{\alpha}u|^{p}\eta^{p\mu}dx\bigg)^{1/p}<\infty, where \eta(x_{3})=\min_{i=1,2}{\rm dist\,}(x_{3},S_{2}(a_{i})). We define also \displaystyle V(\Omega^{t}) \displaystyle\equiv \displaystyle V_{2}^{0}(\Omega^{T})=\{u:||u||_{V^{0}_{2}(\Omega^{T})}={\rm ess}\sup_{t\in(0,T)}||u||_{L_{2}(\Omega)} \displaystyle+\left(\int_{0}^{T}||\nabla u||_{L_{2}(\Omega)}^{2}dt\right)^{1/2}<\infty\}, \displaystyle V_{2}^{1}(\Omega^{T}) \displaystyle\equiv \displaystyle\{u:||u||_{V^{1}_{2}(\Omega^{T})}={\rm ess}\sup_{t\in(0,T)}||u||_{H^{1}(\Omega)} \displaystyle+\left(\int_{0}^{T}||\nabla u||_{H^{1}(\Omega)}^{2}dt\right)^{1/2}<\infty\}. Consider the Neumann problem (see ( 3.7 )) (2.1) \displaystyle{{}\Delta\varphi=-{\rm div}\,b\quad} \displaystyle{{}{\rm in}\ \ \Omega,} \displaystyle{{}\bar{n}\cdot\nabla\varphi=0\quad} \displaystyle{{}{\rm on}\ \ S,} \displaystyle{{}\intop_{\Omega}\varphi dx=0,} \displaystyle{{}\int_{\Omega}{\rm div}\,b=0.} Lemma 2.2 (see [ RZ3 ] , [ RZ4 ] ) . Let {\rm div}\,b\in L_{p,\mu}(\Omega) , p\geq 2 , \mu\in(0,1) . Then the weak solution to ( 2.1 ) satisfies the estimate (2.2) \|\nabla^{2}\varphi\|_{L_{p,\mu}(\Omega)}\leq c\|{\rm div}\,b\|_{L_{p,\mu}(\Omega)}. In what follows we use the anisotropic Sobolev spaces W_{p_{1},p_{2}}^{1}(\Omega) and L_{p_{1},p_{2}}(\Omega) , where p_{1},p_{2}\in[1,\infty] with the norms \|u\|_{W_{p_{1},p_{2}}^{k}(\Omega)}=\bigg[\bigg(\intop_{S_{2}}\sum_{|\alpha|\leq k}|D_{x}^{\alpha}u|^{p_{1}}dS_{2}\bigg)^{p_{2}/p_{1}}dx_{3}\bigg]^{1/p_{2}}, where k\in{\mathbb{N}}\cup\{0\} and W_{p_{1},p_{2}}^{0}(\Omega)=L_{p_{1},p_{2}}(\Omega) . Consider the problem (2.3) \displaystyle{{}\theta_{t}+v\cdot\nabla\theta-\mathchar 1403\relax\Delta\theta=0,} \displaystyle{{}{\rm div}\,v=0,\ \ v\cdot\bar{n}|_{S_{1}}=0,\ \ v\cdot\bar{n}|_{S_{2}}=d,\ \ \bar{n}\cdot\nabla\theta|_{S}=0,} \displaystyle{{}\theta|_{t=0}=\theta_{0}.} Lemma 2.3 . Let \inf_{x\in\Omega}\theta_{0}(x)=\theta_{*}>0 and d_{1}\in L_{2}(0,t;L_{\infty}(S_{2}(-a)). Then solutions to ( 2.3 ) satisfy (2.4) \theta\geq\theta_{*}. Proof. Multiply (\ref{2.3})_{1} by (\theta-\theta_{*})_{-}=\min\{\theta-\theta_{*},0\} and integrate over \Omega . Then we obtain (2.5) \displaystyle{{}\frac{1}{2}\frac{d}{dt}\intop_{\Omega}(\theta-\theta_{*})_{-}^{2}dx+\mathchar 1403\relax|\nabla(\theta-\theta_{*})_{-}|^{2}dx} \displaystyle{{}=-\intop_{\Omega}v\cdot\nabla(\theta-\theta_{*})_{-}(\theta-\theta_{*})_{-}dx} The r.h.s. of ( 2.5 ) equals \displaystyle{{}-\frac{1}{2}\intop_{\Omega}v\cdot\nabla(\theta-\theta_{*})_{-}^{2}dx=-\frac{1}{2}\intop_{S_{2}}v\cdot\bar{n}(\theta-\theta_{*})_{-}^{2}dS_{2}} \displaystyle{{}=\frac{1}{2}\intop_{S_{2}(-a)}d_{1}(\theta-\theta_{*})_{-}^{2}dS_{2}-\frac{1}{2}\intop_{S_{2}(a)}d_{2}(\theta-\theta_{*})_{-}^{2}dS_{2}} Hence, ( 2.5 ) takes the form (2.6) \displaystyle{{}\frac{1}{2}\frac{d}{dt}|(\theta-\theta_{*})_{-}|_{2,\Omega}^{2}+\mathchar 1403\relax|\nabla(\theta-\theta_{*})_{-}|_{2,\Omega}^{2}} \displaystyle{{}\leq\frac{1}{2}\intop_{S_{2}(-a)}d_{1}(\theta-\theta_{*})_{-}^{2}dS_{2}\leq\frac{1}{2}|d_{1}|_{\infty,S_{2}(-a)}\intop_{S_{2}(-a)}(\theta-\theta_{*})_{-}^{2}dS_{2}} \displaystyle{{}\leq\frac{1}{2}|d_{1}|_{\infty,S_{2}(-a)}(\varepsilon|\nabla(\theta-\theta_{*})_{-}|_{2,\Omega}^{2}+\frac{c}{\varepsilon}|(\theta-\theta_{*})_{-}|_{2,\Omega}^{2}),} where we used some interpolation. For \varepsilon sufficiently small we obtain (2.7) \displaystyle{{}\frac{d}{dt}|(\theta-\theta_{*})_{-}|_{2,\Omega}^{2}+\mathchar 1403\relax|\nabla(\theta-\theta_{*})_{-}|_{2,\Omega}^{2}} \displaystyle{{}\leq c|d_{1}|_{\infty,S_{2}(-a)}^{2}|(\theta-\theta_{*})_{-}|_{2,\Omega}^{2}.} Dropping the second term on the l.h.s. of ( 2.7 ) and integrating with respect to time yield (2.8) \displaystyle{{}\intop_{\Omega}(\theta(t)-\theta_{*})_{-}^{2}dx\leq\exp\bigg(c\intop_{0}^{t}|d_{1}(t^{\prime})|_{\infty,S_{2}(-a)}^{2}dt^{\prime}\bigg)\cdot} \displaystyle{{}\quad\cdot\intop_{\Omega}(\theta(0)-\theta_{*})_{-}^{2}dx.} Since \theta(0)>\theta_{*} we have that (\theta(0)-\theta_{*})_{-}=0 so also (\theta(t)-\theta_{*})_{-}=0 . This proves ( 2.4 ) and concludes the proof. ∎ Lemma 2.4 . Let \theta_{*} , \theta^{*} be positive constants such that \theta_{*}<\theta^{*} and \theta_{*}\leq\theta_{0}\leq\theta^{*} . Let d_{1}\in L_{2}({\mathbb{R}}_{+},L_{\infty}(S_{2}(-a))) Then solutions to ( 2.3 ) satisfy (2.9) \theta_{*}\leq\theta(t)\leq\theta^{*}. Proof. Multiply (\ref{2.3})_{1} by \theta^{s-1},s>1 and integrate over \Omega . Then we obtain \displaystyle{{}\frac{1}{s}\frac{d}{dt}|\theta|_{s,\Omega}^{s}+\frac{4\mathchar 1403\relax(s-1)}{s^{2}}\intop_{\Omega}|\nabla\theta^{s/2}|^{2}dx} \displaystyle{{}=-\frac{1}{s}\intop_{\Omega}v\cdot\nabla\theta^{s}dx=-\frac{1}{s}\intop_{S_{2}}d\theta^{s}dS_{2}} \displaystyle{{}\leq\frac{1}{s}|d_{1}|_{\infty,S_{2}(-a)}\intop_{S_{2}(-a)}\theta^{s}dS_{2}} \displaystyle{{}\leq\frac{\varepsilon}{s}\intop_{\Omega}|\nabla\theta^{s/2}|^{2}dx+\frac{c}{\varepsilon s}|d_{1}|_{\infty,S_{2}(-a)}^{2}|\theta|_{s,\Omega}^{s}.} For sufficiently small \varepsilon we get \frac{1}{s}\frac{d}{dt}|\theta|_{s,\Omega}^{s}\leq\frac{c}{s}|d_{1}|_{\infty,S_{2}(-a)}^{2}|\theta|_{s,\Omega}^{s}. Dropping s and integrating with respect to time yield |\theta(t)|_{s,\Omega}^{s}\leq\exp\bigg(c\intop_{0}^{t}|d_{1}(t^{\prime})|_{\infty,S_{2}(-a)}^{2}dt^{\prime}\bigg)|\theta(0)|_{s,\Omega}^{s} Using that |\theta(0)|_{\infty,\Omega}\leq\theta^{*} and passing with s\to\infty the above inequality implies |\theta(t)|_{\infty,\Omega}\leq\theta^{*}. Next we calculate the lower bound for \theta . Multiply (\ref{2.3})_{1} by \theta^{-s} , s>1 , and integrate over \Omega . Then, we have \displaystyle{{}-\frac{1}{s-1}\frac{d}{dt}\intop_{\Omega}\frac{1}{\theta^{s-1}}dx-\frac{4s\mathchar 1403\relax}{(s-1)^{2}}\intop_{\Omega}\bigg|\nabla{1\over\theta^{(s-1)/2}}\bigg|^{2}dx} \displaystyle{{}=\frac{1}{s-1}\intop_{\Omega}v\cdot\nabla\frac{1}{\theta^{s-1}}dx} \displaystyle{{}=-\frac{1}{s-1}\intop_{S_{2}(-a)}d_{1}\frac{1}{\theta^{s-1}}dS_{2}+\frac{1}{s-1}\intop_{S_{2}(a)}d_{2}\frac{1}{\theta^{s-1}}dS_{2}.} Multiplying by -1 we get \displaystyle{{}\frac{1}{s-1}\frac{d}{dt}\intop_{\Omega}\frac{1}{\theta^{s-1}}dx+{4s\mathchar 1403\relax\over(s-1)^{2}}\intop_{\Omega}\bigg|\nabla{1\over\theta^{(s-1)/2}}\bigg|^{2}dx} \displaystyle{{}\leq\frac{1}{s-1}\intop_{S_{2}(-a)}d_{1}\frac{1}{\theta^{s-1}}dS_{2}\leq\frac{1}{s-1}|d_{1}|_{\infty,S_{2}(-a)}\intop_{S_{2}}\frac{1}{\theta^{s-1}}dS_{2}} \displaystyle{{}\leq\frac{\varepsilon}{s-1}\intop_{\Omega}\bigg|\nabla\frac{1}{\theta^{(s-1)/2}}\bigg|^{2}dx+\frac{c}{\varepsilon(s-1)}|d_{1}|_{\infty,S_{2}(-a)}^{2}\intop_{\Omega}\frac{1}{\theta^{s-1}}dS_{2}.} For