References & Citations
Mathematics > Probability
Title: Collision of eigenvalues for matrix-valued processes
(Submitted on 15 Feb 2018 (v1), last revised 9 Jan 2019 (this version, v2))
Abstract: We examine the probability that at least two eigenvalues of an Hermitian matrix-valued Gaussian process, collide. In particular, we determine sharp conditions under which such probability is zero. As an application, we show that the eigenvalues of a real symmetric matrix-valued fractional Brownian motion of Hurst parameter $H$, collide when $H<1/2$ and don't collide when $H>\frac{1}{2}$, while those of a complex Hermitian fractional Brownian motion collide when $H<\frac{1}{3}$ and don't collide when $H>\frac{1}{3}$. Our approach is based on the relation between hitting probabilities for Gaussian processes with the capacity and Hausdorff dimension of measurable sets.
Submission history
From: Arturo Jaramillo [view email][v1] Thu, 15 Feb 2018 05:26:30 GMT (23kb)
[v2] Wed, 9 Jan 2019 09:12:46 GMT (24kb)
Link back to: arXiv, form interface, contact.