Compressed sensing refers to the exact or approximate recovery of very high-dimensional but sparse vectors from a small number of possibly measuremnts. A recent concept is that of "nearly ideal behavior," whereby the estimation error of an algorithm is bounded by a universal constant times the error achievable by an "oracle" that knows the locations of the nonzero components of the vector to be determined. In this talk I will review the basic and well-known results and some recent results, and present some new results that provide a framework for constructing algorithms with near-ideal behavior.Vidyasagar Mathukumalli, 16.05.2014.
