A Transformation of Accelerated Double Step Size Method for Unconstrained Optimization

Abstract EN

A reduction of the originally double step size iteration into the single step length scheme is derivedunder the proposed condition that relates two step lengths in the accelerated double step size gradientdescent scheme.

The proposed transformation is numerically tested.

Obtained results confirm thesubstantial progress in comparison with the single step size accelerated gradient descent methoddefined in a classical way regarding all analyzed characteristics: number of iterations, CPU time,and number of function evaluations.

Linear convergence of derived method has been proved.