Singular variation of nonlinear eigenvalues (Q1201503): Difference between revisions

Let \(M\) be a smoothly bounded domain in \(\mathbb{R}^ 3\), \(w\in M\), \(p\in (1,2)\), \(\varepsilon>0\). Put \(\lambda(M)= \inf\{ \int_ M| \nabla u|^ 2 dx\); \(u\in H_ 0^ 1(M)\), \(\| u\|_{L_{p+1}(M)} =1\}\) and \(M_ \varepsilon= \{x\in M\); \(| x-w|> \varepsilon\}\). Assuming that (i) the positive solution \(u\) of \(-\Delta u=\lambda (M)u^ p\) in \(H_ 0^ 1(M)\) is unique, (ii) the positive minimizer \(u_ \varepsilon\) for \(\lambda(M_ \varepsilon)\) is unique for \(\varepsilon\) small, (iii) \(\text{Ker} (\Delta+ \lambda(M_ \varepsilon) pu_ \varepsilon^{p- 1})= \{0\}\) for \(\varepsilon\) small, the author gives the outline of the proof of the estimate \(\lambda(M_ \varepsilon)- \lambda(M)= 4\pi\varepsilon u(w)^ 2+ o(\varepsilon)\).