Spectral analysis of diffusions with jump boundary (Q639520): Difference between revisions

This article answers in the negative two questions raised by Ben-Ari and Pinsky, namely, whether the eigenvalue of the generator of a diffusion with jump boundary possessing the least real part is necessarily real, and whether its spectral gap is larger than the bottom of the spectrum of the diffusion killed at the boundary. Using the known fact that the former implies the latter, the authors construct a counterexample to the latter, thus disproving both. The counterexample is a scalar diffusion in an interval with constant drift and diffusion and a deterministic jump distribution, for which the aforementioned spectral gap is independent of the drift when the drift is large enough, but the corresponding bottom of the spectrum of the killed diffusion grows unbounded. Both quantities are calculated explicitly.