A non-uniform convergence tolerance scheme for enhancing the branch-and-bound method

Sangjin Jung, Xi Chen, Gyunghyun Choi, Dong Hoon Choi

Research output: Contribution to journalArticlepeer-review


In order to improve the efficiency of the branch-and-bound method for mixed-discrete nonlinear programming, a nonuniform convergence tolerance scheme is proposed for the continuous subproblem optimizations. The suggested scheme assigns the convergence tolerancesfor each continuous subproblem optimization according to the maximum constraint violation obtained from the first iteration of each subproblem optimization in order to reduce thetotal number of function evaluations needed to reach the discrete optimal solution. The proposed tolerance scheme is integrated with five branching order options. The comparative performance test results using the ten combinations of the five branching orders and two convergence tolerance schemes show that the suggested non-uniform convergence tolerance scheme is obviously superior to the uniform one. The results also show that the branching order option using the minimum clearance difference method performed best among the five branching order options. Therefore, we recommend using the "minimum clearance difference method" for branching and the "non-uniform convergence tolerance scheme" for solving discrete optimization problems.

Original languageEnglish (US)
Pages (from-to)361-371
Number of pages11
JournalTransactions of the Korean Society of Mechanical Engineers, A
Issue number4
StatePublished - Apr 1 2012

All Science Journal Classification (ASJC) codes

  • Mechanical Engineering


Dive into the research topics of 'A non-uniform convergence tolerance scheme for enhancing the branch-and-bound method'. Together they form a unique fingerprint.

Cite this