Improving the Hungarian assignment algorithm

Authors:
Roy Jonker;Ton Volgenant
Affiliations:
Department of Operations Research, Faculty of Actuarial Science and Econometrics, University of Amsterdam, 1011 NH Amsterdam, The Netherlands;Department of Operations Research, Faculty of Actuarial Science and Econometrics, University of Amsterdam, 1011 NH Amsterdam, The Netherlands
Venue:
Operations Research Letters
Year:
1986

Citing 2
Cited 3

Algorithm 548: Solution of the Assignment Problem [H]

ACM Transactions on Mathematical Software (TOMS)
An algorithm for the assignment problem

Communications of the ACM

Lower tolerance-based Branch and Bound algorithms for the ATSP

Computers and Operations Research
Iterative patching and the asymmetric traveling salesman problem

Discrete Optimization
A framework for simulation-based optimization of business process models

Simulation

Quantified Score

Hi-index	0.00

Visualization

Abstract

We describe three easily implementable improvement for the Hungarian linear assignment algorithm. Computation times vary from about two to more than three times lower than previously, where the effectiveness increases with problem size. Furthermore, the algorithm is now less sensitive to the range of the cost coefficients. We also show that the Hungarian algorithm is essentially equivalent to assignment algorithms based on shortest augmenting paths.