| Summary: | 碩士 === 國立清華大學 === 數學系 === 85 === Arbel and Oren present the first interion-point multiobjective
linear programming(MOLP) algorithms, whichare based on the
interior-point algorithms for solvingthe usual (single-
objective) linear programs. In this paper, we develop a
variation of Arbel's MOLP primal-dual interior algorithm. Under
the assumption that Decision Makerhas an explicitly known linear
utility function, we prove that ouralgorithm is strictly
increasing in the utility.We apply this algorithm to large MOLP
problems with linear utility functions, which come from the well
known Netlib set test problems. As far as we know, we are the
first to use an interior-point algorithm for solving large MOLP
problems. Since large MOLP are difficult, the numerical results
show that our algorithm in general can not find the optimal
solutions of the utilityprograms to satisfactory accuracies. We
point out the reasons why MOLP are difficult and two possible
improvements for future research.
|
|---|
