Approximation properties of some vector weak biorthogonal greedy algorithms (Q7034053)

This paper targets at the theory of sparse approximation. It extends the framework of Weak Biorthogonal Greedy Algorithms (WBGA) to the multi-target, vector-valued setting in Banach spaces. The authors propose and analyze three new algorithms -- the Vector Weak Chebyshev Greedy Algorithm (VWCGA), the Vector Weak Greedy Algorithm with Free Relaxation (VWGAFR), and the Vector Rescaled Weak Relaxed Greedy Algorithm (VRWRGA) -- demonstrating their efficiency for simultaneous approximation of multiple elements. Through a unified analytical approach, the paper establishes both necessary and sufficient conditions for convergence, as well as optimal convergence rates on certain classes of functions, proving that these algorithms inherit the properties of their scalar counterparts while being exceptionally well-suited for multi-task problems such as signal vector processing and multi-target optimization.