Limited Memory Bundle Method for Large Bound Constrained Nonsmooth Optimization

1. Abstract Practical optimization problems often involve nonsmooth functions of hundreds or thousands of variables. As a rule, the variables in such large problems are restricted to certain meaningful intervals. In the report [Haarala, Mäkelä, 2006] we have described an efficient adaptive limited memory bundle method for large-scale nonsmooth, possibly nonconvex, bound constrained optimization. Although it works very well in numerical experiments it suffers from one theoretical drawback. Namely, it is not necessarily globally convergent. In this paper, a globally convergent variant of this method is proposed. In addition, some results from numerical experiments are given. 2.