The Projected Subgradient Algorithm in Convex Optimization

The discussion takes into consideration the fact that for every algorithm its iteration consists of several steps and that computational errors for different steps are different, in general.