The spectral bundle method with second-order information

C. Helmberg, M. L. Overton, F. Rendl

Research output: Contribution to journalArticlepeer-review


The spectral bundle (SB) method was introduced by Helmberg and Rend [A spectral bundle method for semidefinite programming. SIAM J. Optim. 10 (2000), pp. 673-696] to solve a class of eigenvalue optimization problems that is equivalent to the class of semidefinite programs with the constant trace property. We investigate the feasibility and effectiveness of including full or partial second-order information in the SB method, building on work of Overton [On minimizing the maximum eigenvalue of a symmetric matrix. SIAM J. Matrix Anal. Appl. 9(2) (1988), pp. 256-268] and Overton and Womersley [Second derivatives for optimizing eigenvalues of symmetric matrices. SIAM J. Matrix Anal. Appl. 16 (1995), pp. 697-718]. We propose several variations that include second-order information in the SB method and describe efficient implementations. One of these, namely diagonal scaling based on a low-rank approximation of the second-order model for max, improves the standard SB method both with respect to accuracy requirements and computation time.

Original languageEnglish (US)
Pages (from-to)855-876
Number of pages22
JournalOptimization Methods and Software
Issue number4
StatePublished - Jul 4 2014


  • bundle methods
  • eigenvalue optimization
  • semidefinite optimization

ASJC Scopus subject areas

  • Software
  • Control and Optimization
  • Applied Mathematics


Dive into the research topics of 'The spectral bundle method with second-order information'. Together they form a unique fingerprint.

Cite this