Title:Geometric Perspective on Concentration Phenomena in Frame Theory

Abstract:Parseval and equal-norm frames play a fundamental role in frame theory and signal processing. It is known that a random frame, with unit vectors drawn independently from the uniform distribution on the sphere, will be nearly Parseval with high probability; asymptotic results go back at least to Goyal, Vetterli, and Thao and a non-asymptotic error bound was proved more recently by Kwok, Lau, and Ramachandran. In this work, we prove a dual result, which shows that random Parseval frames, with respect to the Haar measure, are nearly equal-norm with high probability. Our proofs are geometric in nature, and rely on general measure concentration principles in Riemannian manifolds. Using these techniques, we also give a novel probabilistic upper bound for the Paulsen problem.