A nonsmooth version of the univariate optimization algorithm for locating the nearest extremum (locating extremum in nonsmooth univariate optimization) (Q944053)

The author presents an algorithm for solving nonsmooth univariate minimization problems. Finding the step size along the direction vector involves solving a minimization subproblem which is an unidimensional search problem. Hence, unidimensional search methods are most indispensable and the efficiency of any algorithm partly depends on them. In this work an algorithm for univariate optimization using a linear lower bounding function is extended to the nonsmooth case by using the generalized gradient instead of the derivative. A convergence theorem is proved under the condition of semismoothness. This approach gives a globally superlinear convergence of the algorithm, which is a generalized Newton-type method. Four test minimization problems are solved: two with smooth functions and two with nonsmooth ones.