Rate of Convergence of the Bundle Method
- 368 Downloads
The number of iterations needed by the bundle method for nonsmooth optimization to achieve a specified solution accuracy can be bounded by the product of the inverse of the accuracy and its logarithm, if the function is strongly convex. The result is true for the versions of the method with multiple cuts and with cut aggregation.
KeywordsNonsmooth optimization Bundle method
Mathematics Subject Classification90C25
This work was partially supported by the National Science Foundation Award DMS-1312016 and the Air Force Office of Scientific Research Award FA95550-15-1-0251.
- 1.Lemaréchal, C.: Nonsmooth optimization and descent methods, Research Report 78–4. International Institute of Applied Systems Analysis, Laxenburg (1978)Google Scholar