Computational mathematical programming

K. L. Hoffman0444703098, 9780444703095, 9783642009327, 3642009328, 9783642009334, 3642009336

As modelling efforts attempt to solve problems related to ever more complex systems, and as algorithms are developed specifically to handle problems having thousands (or even hundreds of thousands) of variables, the need for sound computational testing and full disclosure of experimental results is both obvious and immediate. This collection of papers reflects both the current technology available to mathematical programmers for solving optimization problems, and mechanisms for testing and determining the quality of software used. This testing requires suitable test problems and the execution of a designed experiment to determine the efficiency, robustness, reliability and applicability of various algorithms. Many aspects of computational mathematical programming are covered: the testing of new algorithms for optimizing functions of a specified form, computational comparisons of known algorithms, a preliminary computational evaluation of the new projective method, why decomposition methods have not been as successful as originally anticipated, etc.