scientific article; zbMATH DE number 3875823

From MaRDI portal
Publication:3343007





In an earlier paper [Arch. Ration. Mech. Anal. 81, 181-197 (1983; Zbl 0516.35031)] we discussed the uniqueness of positive classical solutions u(x) of the problem \((I)\quad\Delta u+f(u)=0\) in \({\mathbb{R}}^ n\), u(x)\(\to 0\) as \(x\to\infty \) in which \(n>1\) and \(x=(x_ 1,...,x_ n)\). Among the basic assumptions on the function f is condition \((H)\lim_{u\to 0}f(u)/u=-m\) for some positive number m. The object of the reviewed paper is to weaken this assumption considerably. It is replaced by \(H(^*)\). One of the following conditions holds: (i) \(n=2\) and \(\beta >0\), (ii) \(n>2\) and \(\alpha >0\). Hypothesis (H) of course implies \(\alpha >0\), so that \((H^*)\) obviously is a considerably weaker condition than (H). In particular, we note that \((H^*)\) allows the possibility \(f\equiv 0\) for u near 0. The main result of the paper is the following theorem: Under certain conditions including \((H^*)\) there can be at most one non- negative (and non-trivial) radial solution \(u=u(r)\) of the problem (I). This solution moreover satisfies the conditions (i) \(u(0)>0\), \(u'(0)=0\), (ii) \(u'(r)<0\) whenever \(r>0\) and \(u>0\), (iii) if \(u(r)=0\) at some value \(r=a\), then \(u(r)=0\) for all \(r\geq a.\) Concerning the possibility that \(u(r)=0\) at some finite value \(r=a\), we also prove the following result: Let \(u=u(r)\) be a non-negative (and non- trivial) radial solution of problem (I). If \(\int_{0}ds/\sqrt{F(x)}<\infty\) then there exists an \(a<\infty\) such that \(u(a)=0\). The converse also holds provided that \(\alpha >0\).











This page was built for publication:

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q3343007)