A set covering formulation of the matrix equipartition problem
- 104 Downloads
This paper is concerned with a certain matrix decomposition problem which has been shown to be NP-hard (MATRIX EQUIPARTITION). Given a (0, 1)-matrix M with row-set R, MATRIX EQUIPARTITION consists in finding two equicardinality subsets R 1 and R 2 of R with maximum size, such that every row of R 1 is disjoint from every row of R 2. In addition to its theoretical significance, the problem arises also in applicative contexts like, for example, the design of Very Large Scale Integrated circuits (VLSI-design) and Flexible Manufacturing Systems (FMS). We prove that MATRIX EQUIPARTITION admits a Set Covering formulation. Although such formulation contains exponentially many constraints, it is easy to check implicitly whether a (0, 1)-vector satisfles all of them and, if not, to generate a set of violated constraints from the formulation. Such property is used to design an incremental algorithm to solve the problem to optimality. We tested the algorithm on several test problems and compared it to a standard Branch & Bound strategy.
KeywordsBoolean Function Flexible Manufacture System Incremental Algorithm Incidence Vector Programme Logic Array
Unable to display preview. Download preview PDF.
- G. De Micheli, M. Santomauro, A clustering algorithm for partitioning of Programmed Logic Arrays, Mem. UCB/ERL M82/74, Electronics Research Laboratory, College of Engineering, University of California, Berkeley, CA (1982).Google Scholar
- H. Fleisher, L.I. Maissel, An introduction to array logic, IBM J. Res. Develop., March 1975, 98–109.Google Scholar
- R.E. Marsten, ZOOM/XMP User’s Manual, Dept. of Management Information Systems, University of Arizona, Tucson, AZ (1987).Google Scholar
- G.L. Nehmhauser, L.A. Wolsey, Integer and combinatorial optimization, J. Wiley&sons, New York 1988.Google Scholar
- P. Nobili, A. Sassano, A. Branch & Cut algorithm for the set covering problem, in preparation.Google Scholar