RESEARCH OF ONE VARIANT OF SUBGRADIENT METHOD

Abstract

The subgradient step selection method based on the known minimal value of function is studied in the paper. The authors show that it is an analogue of the method of minimal errors for solving linear equation systems. For a sequence of the minimum function values on the current set of method iterations, the estimate for the rate of convergence in the form of inequality is received.

Keywords

• subgradient
• convex function
• linear algebra
• minimum of function
• rate of convergence