Convergence of a modified algorithm of fast probabilistic modeling

Abstract

The convergence of fast probabilistic modeling algorithms (G-algorithms) is analyzed. A G-algorithm is modified based on a new probabilistic approach, used to reject points in the neighborhood of the current solution. A theoretically justified estimate of the rate of convergence, independent of the initial approximation, is obtained for this modification. A computational experiment is conducted to compare the performance of the modified G-algorithm with that of the classical one.