Optimal uniqueness theorems and exact blow-up rates of large solutions (Q2493088)

Let \(\Omega\) be a bounded smooth domain in \(\mathbb R^N,\) let \(p>1,\;\lambda\geq 0\) be real numbers, and let \(a(x)> 0\) be a smooth function. The paper deals with strong positive solutions of the boundary blow-up problem \((\Delta +\lambda)u=a(x)u^p\) in \(\Omega\), \(u(x)\to\infty\) as \(x\to\partial\Omega\). The main result concerns the following precise boundary blow-up rate of a solution: if \(x_0\in\partial\Omega\) then \[ \lim_{x\to x_0}\frac{u(x)}{F_{x_0}(\text{dist}(x,\partial\Omega))}=I_0^{-\frac{p}{p-1}} \Bigl(\frac{p+1}{p-1}\Bigr)^\frac{p+1}{p-1}, \] where \[ F_{x_0}(t)=\int_t^\infty\Bigl(\int_0^s\bigl(f_{x_0}(\tau)\bigr)^\frac{1}{p+1}\,d\tau\Bigr) ^{-\frac{p+1}{p-1}},\;\;\;I_0=\lim_{t\to 0}\frac{F_{x_0}(t)F''_{x_0}(t)}{[F'_{x_0}(t)]^2}, \] and \(f_{x_0}(t)\) is any smooth extension of the boundary normal section of \(a(x)\) at \(x_0\), that is, \(f_{x_0}(t)=a(x_0-t{\mathbf n}_{x_0},)\) \(t>0\), \(t\) small. As usual, \({\mathbf n}_{x_0}\) stands for the outward unit normal of \(\partial\Omega\) at \(x_0\). The function \(a(x)\) is assumed to satisfy suitable requirements, in particular, the corresponding \(f_{x_0}(t)\) belongs to \(L^\frac{1}{p+1}.\) Note that, in case of \(a(x)=1\), the previous result was well known. The paper is organized as follows. First, assuming \(f\in C[0,\infty)\) satisfies \(0<f(t)\leq f(s)\) for \(0<t<s,\) the existence and uniqueness of the one dimensional problem \[ u''=fu^p,\;\;t>0,\;\;u(0)=\infty,\;\;u(\infty)=0, \] is proved. Next, if \(f\) satisfies suitable additional assumptions, the exact blow-up rate of the solution \(u(t)\) near \(t=0\) in terms of the function \[ F(t)=\int_t^\infty\Bigl(\int_0^s\bigl(f(\tau)\bigr)^\frac{1}{p+1}\,d\tau\Bigr) ^{-\frac{p+1}{p-1}} \] is found. By using this result, the boundary blow-up problem in the case of balls and annulus is discussed. Finally, by using the comparison principle for elliptic equations, the problem in a general domain is investigated. The boundary estimate (which has a own interest) is used to prove uniqueness of the boundary blow-up problem.