The Projected Subgradient Method for Nonsmooth Convex Optimization in the Presence of Computational Errors.

We study the convergence of the projected subgradient method for constrained convex optimization in a Hilbert space. Our goal is to obtain an ε-approximate solution of the problem in the presence of computational errors, where ε is a given positive number. The results that we obtain are important in practice because computations always introduce numerical errors. [ABSTRACT FROM AUTHOR]