Improved Penalties for Fixed Cost Linear Programs Using Lagrangean Relaxation
The most commonly used penalty in branch and bound approaches to integer programming is the Driebeek--Tomlin penalty. It has been used successfully in solving fixed cost linear programs by Kennington and Unger and by Barr, Glover and Klingman. It is well known that the Driebeek--Tomlin penalty can be derived from a Lagrangean relaxation of the integer programming problem. We show, however, that the Lagrangean relaxation for fixed cost problems not only yields the Driebeek--Tomlin penalty, but two penalties which sometimes dominate it. We show the strength of the new penalties by solving a series of text problems and comparing the number of nodes generated on the branch and bound tree and the total computer time needed to solve each problem.
Volume (Year): 32 (1986)
Issue (Month): 7 (July)
|Contact details of provider:|| Postal: |
Web page: http://www.informs.org/Email:
More information through EDIRC
When requesting a correction, please mention this item's handle: RePEc:inm:ormnsc:v:32:y:1986:i:7:p:856-869. See general information about how to correct material in RePEc.
For technical questions regarding this item, or to correct its authors, title, abstract, bibliographic or download information, contact: (Mirko Janc)
If references are entirely missing, you can add them using this form.