An inexact bundle algorithm for nonconvex nonsmooth minimization in Hilbert space

Lukas Hertlein, Michael Ulbrich

Research output: Contribution to journalArticlepeer-review

11 Scopus citations

Abstract

In this paper a bundle method for nonconvex nonsmooth optimization in infinite-dimensional Hilbert spaces is developed and analyzed. The algorithm requires only inexact function value and subgradient information. Global convergence to approximately stationary points is proved, where the final accuracy depends on the error level in the function and subgradient data. The method is then applied to an optimal control problem governed by the obstacle problem. For adaptively controlling the inexactness, implementable conditions are developed, first on a general level and then for the concrete case of a FEM discretization for optimal control of an obstacle problem. Numerical results are presented.

Original languageEnglish
Pages (from-to)3137-3165
Number of pages29
JournalSIAM Journal on Control and Optimization
Volume57
Issue number5
DOIs
StatePublished - 2019

Keywords

  • Error estimates
  • Inexact function values
  • Inexact subgradients
  • Nonconvex bundle method
  • Nonsmooth optimization
  • Optimal control of obstacle problem

Fingerprint

Dive into the research topics of 'An inexact bundle algorithm for nonconvex nonsmooth minimization in Hilbert space'. Together they form a unique fingerprint.

Cite this