On the Generation of Alternative Solutions for Discrete Optimization Problems with Uncertain Data - An Experimental Analysis of the Penalty Method

by    J. Sameith

Preprint series: 04-01 , Reports on Optimization

90C31 Sensitivity, stability, parametric optimization
90B50 Management decision making, including multiple objectives [See also 90C31, 91A35, 91B06]
90C27 Combinatorial optimization

Abstract: The penalty method is a method to generate alternative solutions for many discrete optimization problems. A penalty parameter easily allows to have influence on the difference between optimal and alternative solution.
We experimentally test the penalty method for three optimization problems (shortest paths, assignments, travelling salesman) with parameters that change between planning phase and realization phase.

Keywords: multiple choice system, true alternative, uncertainty, penalty method, k-best algorithm

Upload: 2004-02-09

The author(s) agree, that this abstract may be stored as full text and distributed as such by abstracting services.