Sharp observation time for wave equations with an inverse-square boundary potential

We determine the sharp observation-time threshold for the autonomous, pure-potential wave equation with an inverse-square boundary singularity on the unit ball. For every fixed $κ<0$, full-boundary observability holds on $(-T,T)$ whenever $T>1$ and fails when $T<1$. This resolves, for this model, the optimal-time question raised by Enciso, Shao, and Vergara [J. Eur. Math. Soc. 23 (2021), 3459–3495] and extends their parameter range $-\frac12<κ<0$ to all $κ<0$, including the limit-point regime $κ\leq-\frac12$. Our approach is based on the Ingham–Beurling theory of nonharmonic Fourier series. Spherical harmonic decomposition reduces the problem to one-dimensional singular operators, for which we establish frequency separation, local counting estimates, and lower bounds for the boundary norming constants, uniformly in the angular parameter. A quantitative Ingham–Beurling inequality then yields the sharp-time observability estimate. As a by-product, this spectral approach provides an independent proof of boundary observability and an explicit functional characterization of its maximal $L^2$ observation domain. The latter is described by a weighted $\ell^2$ condition on the spectral coefficients and consists precisely of energy solutions whose distributional Neumann trace belongs to $L^2$. It contains the boundary-admissible energy solutions considered in Enciso–Shao–Vergara's original work.