Uniform anti-maximum principles (Q1577657)

This paper is concerned with some uniform anti-maximum principles for classes of higher-order elliptic boundary values problems \({\mathcal A}u=\lambda u+f\) in \(\Omega\) with homogeneous boundary conditions \({\mathcal B}u= 0\) on \(\partial\Omega\), where \(\Omega\) is a bounded domain in \(\mathbb{R}^n\). The authors prove that, if there exists a simple eigenvalue \(\lambda_1\) for which the corresponding eigenfunction is positive and satisfies appropriate boundary estimates, then for positive \(f\in L^p(\Omega)\) with \(p\) large enough there exists \(\delta_f> 0\) such that the solution \(u\) is negative for \(\lambda\in (\lambda_1,\lambda_1+ \delta_f)\) and positive for \(\lambda\in (\lambda_1- \delta_f, \lambda_1)\). This property is called ``an anti-maximum principle''. Also, they give conditions such that this sign reserving property is uniform, that is, there exists \(\delta> 0\) such that for all positive \(f\) the solution \(u\) is negative for \(\lambda\in (\lambda_1,\lambda_1+ \delta)\) and positive for \(\lambda\in (\lambda_1- \delta,\lambda_1)\). This property is called ``a uniform anti-maximum principle''. A typical example of \({\mathcal A}\) is \({\mathcal A}= (-\Delta)^m\).