Abstract: "A new technique called balancing is presented for the solution of Knapsack Problems. It is proved that an optimal solution to KP is balanced, and thus only balanced feasible solutions need to be enumerated in order to solve the problem to optimality. Restricting a dynamic programming...

Ausführliche Beschreibung

Bibliographische Detailangaben

1. Verfasser:	Pisinger, David (VerfasserIn)
Format:	Buch
Sprache:	English
Veröffentlicht:	København 1995
Schriftenreihe:	Datalogisk Institut <København>: DIKU-Rapport 1995,6
Schlagworte:	Computational complexity Dynamic programming Integer programming NP-complete problems
Zusammenfassung:	Abstract: "A new technique called balancing is presented for the solution of Knapsack Problems. It is proved that an optimal solution to KP is balanced, and thus only balanced feasible solutions need to be enumerated in order to solve the problem to optimality. Restricting a dynamic programming algorithm to only consider balanced states, implies that the Subset-sum Problem, 0-1 knapsack Problem, Multiple-choice Subset- sum Problem and Bounded Knapsack Problem all are solvable in linear time provided that the coefficients are bounded by a constant. Extensive computational experiments are presented to document that the derived algorithms also in practice solve several difficult problems from the literature faster than previous approaches. It is discussed how the presented enumeration schemes may affect approximate algorithms for the problems considered, and how balancing may be combined with other techniques known for Knapsack Problems in order to obtain tighter time bounds."
Beschreibung:	12 S.