The behaviour, of stationary solutions to some nonlinear diffusion problems

From MaRDI portal





The objective of this paper is to study the positive solutions to the boundary value problem \[ -\Delta_pu=\lambda f(u),\quad x\in\Omega,\qquad u=0,\quad x\in\partial\Omega, \] for large \(\lambda\). The operator \(\Delta_p\) stands for the \(p\)-Laplacian, i. e. \(\Delta_p:=\text{ div }\left( |\nabla|^{p-2}\nabla u\right)\) where \(p>1\). A solution \(u\) is an element \(u\in W_0^{1,p}(\Omega)\cap L^\infty(\Omega)\) and is understood in a week sense. The main result is: \smallskip Theorem. Let \(\Omega\subset \mathbb{R}^N\) be a bounded \(C^{2+\beta}\) \((0<\beta\leq 1)\) domain and let \(f=f(u)\) be a continuous function, \(f(u)=u^{p-1}h(u)\) with \(p>1\) and \(h=h(u)\) being a \(C^1\) decreasing function such that \(0<h(0+)=\sigma<+\infty\) and \(h(u)\) having a positive zero \(u=u_0\) of order \(k\). For \(\lambda>\lambda_{1,p}\) let \(u=u_\lambda(x)\) be the unique positive solution of the considered problem, which satisfies \(0<u_\lambda(x)\leq u_0\) for each \(x\in\Omega\) (there is a reference to previous results claiming that there is \(\lambda_{1,p}\) such that the unique positive solution of the problem exists for \(\lambda>\lambda_{1,p}/\sigma\) and does not exist for \(\lambda<\lambda_{1,p}/\sigma\)). Then, i) The family of solutions \(u_\lambda\) satisfies \(\lim_{\lambda\to+\infty}u_\lambda(x)=u_0\), ii) Moreover, the precise estimate of the boundary layer is given by \[ \lim_{\lambda\to+\infty}\lambda^{-1/p} {\partial\over\partial n}u_\lambda(x)= -\left({p\over p-1}\right)^{1/p} (F(u_0))^ {1/p}, \] uniformly in \(\partial\Omega\), where \(F(u_0)=\int_0^{u_0}f(s) ds\).











This page was built for publication: The behaviour, of stationary solutions to some nonlinear diffusion problems

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4378815)