Title:  Linearized Krylov subspace Bregman iteration with nonnegativity constraint

Abstract: Bregman-type iterative methods have attracted considerable attention

in recent years due to their ease of implementation and the high quality of the

computed solutions they deliver. However, these iterative methods may 

require alarge number of iterations and this reduces their attractiveness. This talk

describes a linearized Bregman algorithm defined by projecting the 

problem tobe solved into an appropriately chosen low-dimensional Krylov subspace. The projection reduces both the number of iterations and the computational  effort required for each iteration. A variant of this solution method, in which nonnegativity of each computed iterate is imposed, also is described. 

The talk presents joint work with A. Buccini and M. Pasha.