An improved convergence result for the discrete gradient and secant methods for nonsmooth optimization.

6 pages ; 21 cm

Bibliography p. 6

Bibliografia s. 6

The article deals with generalization of the non-derivative discrete gradient method of Bagirov et al. for minimizing a locally Lipschitz function f on Rn . The existing convergence result for this method has been strengthened by showing that it either drives the f -values to -∞ or each of its cluster points is Clarke stationary for f, without requir­ing compactness of the level sets of f. This generalization is an approximate bundle method, which also subsumes the secant method of Bagirov et al.

6 stron ; 21 cm