Abstract
This paper establishes a sufficient condition for a variable of a linear program to be positive at all optimal solutions. A numerical test using the condition is incorporated into Karmarkar's new LP algorithm to determine columns of optimal basis. Experimental results on the test are also reported.
Similar content being viewed by others
References
V. Chvatal,Linear Programming, Freedman, New York, 1983.
N. Karmarkar, A new polynomial-time algorithm for linear programming,Proceedings of the 16th Annual ACM Symposium on Theory of Computing, Washington, D.C., 1984.
N. Karmarkar, A new polynomial-time algorithm for linear programming,Combinatorica,4 (1984), 373–395.
N. Megiddo, A variation on Karmarkar's algorithm, Preliminary Report, IBM Research Laboratory, San Jose, CA, 1985.
M. J. Todd and B. P. Burrell, An extension of Karmarkar's algorithm for linear programming using dual variables, Technical Report No. 648, School of Operations Research and Industrial Engineering, Cornell University, Ithaca, NY, 1985.
J. A. Tomlin, An experimental approach to Karmarkar's linear programming algorithm, Manuscript, Ketron Inc., Mountain View, CA, 1985.
K. Tone, A hybrid method for linear programming, Research Report 85-B-l, Institute for Policy Science, Saitama University, Urawa, Saitama 338, 1985.
Author information
Authors and Affiliations
Additional information
Communicated by Nimrod Megiddo.
Rights and permissions
About this article
Cite this article
Kojima, M. Determining basic variables of optimal solutions in Karmarkar's new LP algorithm. Algorithmica 1, 499–515 (1986). https://doi.org/10.1007/BF01840459
Received:
Revised:
Issue Date:
DOI: https://doi.org/10.1007/BF01840459