Random sections of p-ellipsoids and optimal recovery

Description

Understanding whether the circumradius of a random section of a set is close to the minimal circum- radius is related to understanding the power of random (Gaussian) information compared to optimal information for recovering vectors from this set. We consider p-ellipsoids, which are generalizations of ellipsoids, and present recent results on the power of random information for them. In the case of poly- nomially decaying semiaxes, we find that for a certain range of parameters we have a threshold of decay, above of which Gaussian information is close to optimal and below of which it is useless. Concepts from compressed sensing will be used for an upper bound for non-convex p-ellipsoids. This is joint work with A. Hinrichs and J. Prochno.

Period	10 Sept 2021
Event title	9th Austrian Stochastics Days, September 9-10, 2021, Leoben
Event type	Conference
Location	AustriaShow on map