Characterization of the Tau Spectrum for Eigenvalue Problems

We consider tau approximations of a general eigenvalue problem for a K-th-order linear differential operator in one dimension, and we show that the approximate spectrum satisfies a resolvent-based spectral determinant property, which we use to investigate it. More specifically, we analyze the spectral determinant property for a second-order Sturm-Liouville eigenvalue problem approximated using Gegenbauer tau methods. Notably, a quantitative error analysis identifies two values of the Gegenbauer parameter $γ$ leading to better approximation properties. These values correspond to Legendre polynomials and their derivatives. Moreover, we discuss the behavior of the largest eigenvalue of the tau approximation. The results are validated with a numerical implementation.