Convergence control in congested assignment is an important, but generally neglected task. The term convergence refers to the degree to which the iterative solution process has achieved the assignment objective. This research project investigates the way in which convergence is currently monitored in practice, and how to improve this quality control in terms of convergence indicator used and minimum stop values required. The research has followed four tracks: (1) a literature study; (2) discussions with international assignment experts; (3) a survey of English users of commercial assignment packages; and (4) model simulations. Results indicate that in practice convergence is not often explicitly monitored, and that even then generally default convergence criteria and associated stop values are used. Convergence should be monitored with respect to both stability of the assignment results and proximity to the assignment objective. (A)
Abstract