Symmetry minimizes the principal eigenvalue: an example for the Pucci's sup-operator (Q2255331)

Consider the supremum Pucci operator \[ M_{\lambda,\Lambda}^+ (X) = \lambda \sum_{e_i<0} e_i + \Lambda \sum_{e_i>0}e_i, \] where the \(e_i\)'s are the eigenvalues of the symmetric matrix \(X\) and \(\Lambda \geq \lambda > 0\) are fixed constants. It is known that if \(\Omega\) is a bounded Lipschitz domain, then there exists \(\phi > 0\) in \(\Omega\) such that \[ M_{\lambda,\Lambda}^+ (D^2\phi) + \mu^+(\Omega) \phi = 0\quad \text{in } \Omega \] with \(\phi = 0\) on \(\partial \Omega\) for \[ \mu^+(\Omega) = \sup \{ \mu \in \mathbb{R} : \exists \phi > 0 \text{ in } \Omega,\, M_{\lambda,\Lambda}^+ (D^2\phi) + \mu(\Omega) \phi \leq 0 \text{ in } \Omega \} \] and similarly for the existence of a negative eigenfunction for \(\mu^-(\Omega)\). Aiming at the natural question whether the Faber-Krahn inequality is valid for these eigenvalues associated with \(M_{\lambda,\Lambda}^+\), the authors show that among a family of subsets of \(\mathbb{R}^2\) with the same area, which are in some sense deformations of rectangles, the one that minimizes \(\mu^+(\,\cdot\,)\) is the most symmetric one. This minimal domain is denoted by \(\Omega_1^\omega\) for \(\omega = \Lambda / \lambda\). The result is accomplished by explicitly computing the eigenvalue \(\mu^+(\Omega_1^\omega)\) and the corresponding eigenfunction.