The Bernstein Branch-and-Prune Algorithm for Constrained Global Optimization of Multivariate Polynomial MINLPs
This paper address the global optimization problem of polynomial mixed-integer nonlinear programs (MINLPs). A improved branch-and-prune algorithm based on the Bernstein form is proposed to solve such MINLPs. The algorithm use a new pruning feature based on the Bernstein form, called the Bernstein box and Bernstein hull consistency. The proposed algorithm is tested on a set of 16 MINLPs chosen from the literature. The efficacy of the proposed algorithm is brought out via numerical studies with the previously reported Bernstein algorithms and several state-of-the-art MINLP solvers.
KeywordsTest Problem Equality Constraint MINLP Problem Consistency Technique Constraint Polynomial
This work was funded by the Singapore National Research Foundation (NRF) under its Campus for Research Excellence And Technological Enterprise (CREATE) programme and the Cambridge Centre for Advanced Research in Energy Efficiency in Singapore (CARES).
- 1.The Mathworks Inc., MATLAB version 7.1 (R14), Natick, MA (2005)Google Scholar
- 7.Kuipers, K.: Branch-and-bound solver for mixed-integer nonlinear optimization problems. MATLAB Central for File Exchange. Accessed 18 Dec. 2009Google Scholar
- 8.GAMS Minlp Model Library: http://www.gamsworld.org/minlp/minlplib/minlpstat.htm. Accessed 20 March 2015
- 10.NEOS server for optimization.: http://www.neos-server.org/neos/solvers/index.html. Accessed 20 March 2015