Analysis of Lagrangian lower bounds for a graph partitioning problem
Adil, G. K.
Ghosh, J. B.
0030-364X (print)1526-5463 (online)
Institute for Operations Research and the Management Sciences (INFORMS)
785 - 788
MetadataShow full item record
Please cite this item using this persistent URLhttp://hdl.handle.net/11693/25215
Recently, Ahmadi and Tang (1991) demonstrated how various manufacturing problems can be modeled and solved as graph partitioning problems. They use Lagrangian relaxation of two different mixed integer programming formulations to obtain both heuristic solutions and lower bounds on optimal solution values. In this note, we point to certain inconsistencies in the reported results. Among other things, we show analytically that the first bound proposed is trivial (i.e., it can never have a value greater than zero) while the second is also trivial for certain sparse graphs. We also present limited empirical results on the behavior of this second bound as a function of graph density.