A linear programming based analysis of the CP-rank of completely positive matrices

Li, Yingbo ; Kummert, Anton ; Frommer, Andreas

International Journal of Applied Mathematics and Computer Science, Tome 14 (2004), p. 25-31 / Harvested from The Polish Digital Mathematics Library

Résumé

A real matrix A is said to be completely positive (CP) if it can be decomposed as A = BB^T, where the real matrix B has exclusively non-negative entries. Let k be the rank of A and Φ_k the least possible number of columns of the matrix B, the so-called completely positive rank (cp-rank) of A. The present work is devoted to a study of a general upper bound for the cp-rank of an arbitrary completely positive matrix A and its dependence on the ordinary rank k. This general upper bound of the cp-rank has been proved to be at most k(k + 1)/2. In a recent pioneering work of Barioli and Berman it was slightly reduced by one, which means that Φ_k ≤ k(k + 1)/2 - 1 holds for k ≥ 2. An alternative constructive proof of the same result is given in the present paper based on the properties of the simplex algorithm known from linear programming. Our proof illuminates complete positivity from a different point of view. Discussions concerning dual cones are not needed here. In addition to that, the proof is of constructive nature, i.e. starting from an arbitrary decomposition A = B_1B^T_1 (B_1 ≥ 0) a new decomposition A = B_2B^T_2 (B_2 ≥ 0) can be generated in a constructive manner, where the number of column vectors of B_2 does not exceed k(k + 1)/2 − 1. This algorithm is based mainly on the well-known techniques stemming from linear programming, where the pivot step of the simplex algorithm plays a key role.

Détail

Publié le : 2004-01-01

Zbl 1171.90471

EUDML-ID : urn:eudml:doc:207675

Bibliographie

[000] Barioli F. and Berman A. (2003): The maximal cp-rank of rank completely positive matrices. - Linear Algebra Appl., Vol. 363, pp. 17-33. | Zbl 1042.15012

[001] Berman A. and Plemmons R. (1979): Nonnegative Matrices in the Mathematical Sciences. - New York: Academic Press. | Zbl 0484.15016

[002] Berman A. and Shaked-Monderer N. (2003): Completely Positive Matrices. - New York: World Scientific. | Zbl 1030.15022

[003] Berman A. (1993): Completely positive graphs, In: Combinatorial and Graph-Theoretical Problems in Linear Algebra (R.A. Brnaldi, S. Friedland and V. Klee, Eds.). - New York: Springer, pp. 229-233. | Zbl 0792.05095

[004] Drew J.H., Johnson C.R. and Loewy R. (1994): Completely positive matrices associated with M-matrices. - Linear and Multilinear Algebra, Vol. 37, No. 4, pp.303-310. | Zbl 0815.15019

[005] Drew J.H. and Johnson C.R. (1996): The no long odd cycle theorem for completely positive matrices, In: Random Discrete Structures (D. Aldous and R. Premantle, Eds.). - New York: Springer, pp. 103-115. | Zbl 0838.15011

[006] Golub G.H. and Van Loan Ch.F. (1989): Matrix Computations. - Baltimore: J. Hopkins University Press.

[007] Glashoff K. and Gustafson S. (1983): Linear Optimization and Approximation. - New York: Springer. | Zbl 0526.90058

[008] Gray L.J. and Wilson D.G. (1980): Nonnegative factorization of positive semidefinite nonnegative matrices. - Linear Algebra Appl., Vol. 31. | Zbl 0434.15012

[009] Hall Jr. M. and Newman M. (1963): Copositive and completely positive quadratic forms. - Proc. Cambridge Philos. Soc., Vol. 59, pp. 329-339. | Zbl 0124.25302

[010] Hall Jr. M. (1967): Combinatorial Theory. - Lexington: Blaisdell Hanna J. and Laffey T.J. (1983): Nonnegative factorization of completely positive matrices. - Linear Algebra Appl., Vol. 55, pp. 1-9.

[011] Karloff H. (1991): Linear Programming. - Boston: Birkh Kaykobad M. (1988): On nonnegative factorization of matrices. - Linear Algebra Appl., Vol. 96, pp. 27-33.

[012] Kelly C. (1994): A test of the markovian model of DNA evolution. - Biometrics, Vol. 50, No. 3, pp. 653-664. | Zbl 0822.62095

[013] Kogan N. and Berman A. (1993): Characterization of completely positive graphs. - Discrete Math., Vol. 114, No. 1-3, pp. 297-304. | Zbl 0783.05071