Non-resonance conditions for semilinear Sturm-Liouville problems with jumping nonlinearities (Q5929883): Difference between revisions

The author studies the solvability, in the usual Sobolev space \(H= \text{H}^2((0,\pi), \mathbb{R})\), of the boundary value problem consisting of the generalized Sturm-Liouville equation \[ -(p(x) u'(x))'+ q(x) u(x)= f(x, u(x))+ h(x)\quad\text{on }(0,\pi) \] and a separated real boundary condition such as the Dirichlet boundary condition \(u(0)= 0= u(\pi)\), with \(p\in C^1([0,\pi], \mathbb{R})\), \(p> 0\) on \([0,\pi]\), \(q\in C^0([0, \pi],\mathbb{R})\), \(f: (0,\pi)\times \mathbb{R}\to \mathbb{R}\) such that \(f(x,y)\) is measurable in \(x\) for every fixed \(y\in \mathbb{R}\) and continuous in \(y\) for almost every \(x\in (0,\pi)\), and \(h\in L((0,\pi), \mathbb{R})\). It is assumed that the rate of growth of \(f(x,y)\) is at most linear as \(|y|\to \infty\), but the asymptotic behavior may be different as \(y\to\pm\infty\), and such nonlinearity is termed ``jump''. Usually, solvability conditions are given in terms of the Fucik spectrum of the associated boundary value problem consisting of \[ (pu')'+ qu= \alpha u^+- \beta u^-\quad\text{on }(0,\pi) \] and the same boundary condition, with \(u^\pm(x)= \max\{\pm u(x), 0\}\). Recall that the Fucik spectrum is the set of points \((\alpha,\beta)\in \mathbb{R}^2\) for which this problem has a nontrivial solution in \(H\). The author uses different spectra, i.e., the spectra of the boundary value problems consisting of \[ (pu')'+ qu= vu^+- wu^-+\lambda u\quad\text{on }(0,\pi) \] with \(v,w\in L^\infty((0,\pi), \mathbb{R})\) and the same boundary condition. It has been shown that such a spectrum consists of a sequence of numbers \[ \lambda^{v,m}_{1,l}\leq \lambda^{v,w}_{1,r}< \lambda^{v,w}_{2,l}\leq \lambda^{v,w}_{2,r}<\cdots\;. \] Define four \(L^\infty((0,\pi), \mathbb{R})\)-functions by \[ i_\pm(x)= \liminf_{y\to \pm\infty} {f(x,y)\over y},\quad s_\pm(x)= \limsup_{y\to \pm\infty} {f(x,y)\over y},\quad x\in (0,\pi). \] Then, the main results of this paper can be stated as follows: i) if \(0\in (\lambda^{i_+,i_-}_{k,r}, \lambda^{i_+,i_-}_{k+ 1,l})\cap (\lambda^{s_+,s_-}_{k,r}, \lambda^{s_+, s_-}_{k+1,l})\) for some \(k\geq 0\), then the original boundary value problem has a solution in \(H\); ii) if \(0\in (\lambda^{i_+, i_-}_{k,l}, \lambda^{i_+, i_-}_{k,r})\cap (\lambda^{s_=, s_-}_{k,l}, \lambda^{s_+,s_-}_{k,r})\) for some \(k\geq 1\), then there exists an \(h\in L((0,\pi), \mathbb{R})\) such that the original boundary value problem has no solution in \(H\).