Domain reduction techniques for global NLP and MINLP optimization

摘要

Optimization solvers routinely utilize presolve techniques, including modelsimplification, reformulation and domain reduction techniques. Domain reductiontechniques are especially important in speeding up convergence to the globaloptimum for challenging nonconvex nonlinear programming (NLP) and mixed-integernonlinear programming (MINLP) optimization problems. In this work, we surveythe various techniques used for domain reduction of NLP and MINLP optimizationproblems. We also present a computational analysis of the impact of thesetechniques on the performance of various widely available global solvers on acollection of 1740 test problems.