×

Positive definite matrices and the S-divergence. (English) Zbl 1338.15046

Summary: Hermitian positive definite (hpd) matrices form a self-dual convex cone whose interior is a Riemannian manifold of nonpositive curvature. The manifold view comes with a natural distance function but the conic view does not. Thus, drawing motivation from convex optimization we introduce the S-divergence, a distance-like function on the cone of hpd matrices. We study basic properties of the S-divergence and explore its connections to the Riemannian distance. In particular, we show that (i) its square-root is a distance, and (ii) it exhibits numerous nonpositive-curvature-like properties.

MSC:

15A45 Miscellaneous inequalities involving matrices
52A99 General convexity
47B65 Positive linear operators and order-bounded operators
65F60 Numerical computation of matrix exponential and similar matrix functions
PDFBibTeX XMLCite
Full Text: DOI

References:

[1] Ando, T., Concavity of certain maps on positive definite matrices and applications to Hadamard products, Linear Algebra Appl., 26, 203-241 (1979) · Zbl 0495.15018 · doi:10.1016/0024-3795(79)90179-4
[2] Banerjee, Arindam; Merugu, Srujana; Dhillon, Inderjit; Ghosh, Joydeep, Clustering with Bregman divergences. Proceedings of the Fourth SIAM International Conference on Data Mining, 234-245 (2004), SIAM, Philadelphia, PA
[3] Berg, Christian; Christensen, Jens Peter Reus; Ressel, Paul, Harmonic analysis on semigroups, Graduate Texts in Mathematics 100, x+289 pp. (1984), Springer-Verlag, New York · Zbl 0619.43001 · doi:10.1007/978-1-4612-1128-0
[4] Bhagwat, K. V.; Subramanian, R., Inequalities between means of positive operators, Math. Proc. Cambridge Philos. Soc., 83, 3, 393-401 (1978) · Zbl 0375.47017
[5] Bhatia, Rajendra, Matrix analysis, Graduate Texts in Mathematics 169, xii+347 pp. (1997), Springer-Verlag, New York · Zbl 0863.15001 · doi:10.1007/978-1-4612-0653-8
[6] Bhatia, Rajendra, Positive definite matrices, Princeton Series in Applied Mathematics, x+254 pp. (2007), Princeton University Press, Princeton, NJ · Zbl 1125.15300
[7] Bhatia, Rajendra; Holbrook, John, Riemannian geometry and matrix geometric means, Linear Algebra Appl., 413, 2-3, 594-618 (2006) · Zbl 1088.15022 · doi:10.1016/j.laa.2005.08.025
[8] Bhattacharyya, A., On a measure of divergence between two statistical populations defined by their probability distributions, Bull. Calcutta Math. Soc., 35, 99-109 (1943) · Zbl 0063.00364
[9] Bougerol, Philippe, Kalman filtering with random coefficients and contractions, SIAM J. Control Optim., 31, 4, 942-959 (1993) · Zbl 0785.93040 · doi:10.1137/0331041
[10] Boyd, Stephen; Vandenberghe, Lieven, Convex optimization, xiv+716 pp. (2004), Cambridge University Press, Cambridge · Zbl 1058.90049 · doi:10.1017/CBO9780511804441
[11] Bridson, Martin R.; Haefliger, Andr{\'e}, Metric spaces of non-positive curvature, Grundlehren der Mathematischen Wissenschaften [Fundamental Principles of Mathematical Sciences] 319, xxii+643 pp. (1999), Springer-Verlag, Berlin · Zbl 0988.53001 · doi:10.1007/978-3-662-12494-9
[12] Chebbi, Zeineb; Moakher, Maher, Means of Hermitian positive-definite matrices based on the log-determinant \(\alpha \)-divergence function, Linear Algebra Appl., 436, 7, 1872-1889 (2012) · Zbl 1236.15060 · doi:10.1016/j.laa.2011.12.003
[13] Chen, Pengwen; Chen, Yunmei; Rao, Murali, Metrics defined by Bregman divergences, Commun. Math. Sci., 6, 4, 915-926 (2008) · Zbl 1163.26320
[14] [iccv11] A. Cherian, S. Sra, A. Banerjee, and N. Papanikolopoulos, Efficient Similarity Search for Covariance Matrices via the Jensen-Bregman LogDet Divergence, Int. Conf. Computer Vision (ICCV), Nov. 2011, pp. 2399-2406.
[15] Cover, Thomas M.; Thomas, Joy A., Elements of information theory, Wiley Series in Telecommunications, xxiv+542 pp. (1991), John Wiley & Sons, Inc., New York · Zbl 0762.94001 · doi:10.1002/0471200611
[16] Fiedler, Miroslav, Bounds for the determinant of the sum of hermitian matrices, Proc. Amer. Math. Soc., 30, 27-31 (1971) · Zbl 0277.15010
[17] Gindikin, S. G., Invariant generalized functions in homogeneous domains, Funkcional. Anal. i Prilo\v zen., 9, 1, 56-58, English transl., Functional Anal. Appl. 9 (1975), no. 1, 50-52 (1975)
[18] Kubo, Fumio; Ando, Tsuyoshi, Means of positive linear operators, Math. Ann., 246, 3, 205-224 (1979/80) · Zbl 0412.47013 · doi:10.1007/BF01371042
[19] Lee, Hosoo; Lim, Yongdo, Invariant metrics, contractions and nonlinear matrix equations, Nonlinearity, 21, 4, 857-878 (2008) · Zbl 1153.15020 · doi:10.1088/0951-7715/21/4/011
[20] Lemmens, Bas; Nussbaum, Roger, Nonlinear Perron-Frobenius theory, Cambridge Tracts in Mathematics 189, xii+323 pp. (2012), Cambridge University Press, Cambridge · Zbl 1246.47001 · doi:10.1017/CBO9781139026079
[21] Nesterov, Yurii; Nemirovskii, Arkadii, Interior-point polynomial algorithms in convex programming, SIAM Studies in Applied Mathematics 13, x+405 pp. (1994), Society for Industrial and Applied Mathematics (SIAM), Philadelphia, PA · Zbl 0824.90112 · doi:10.1137/1.9781611970791
[22] [ssdiv] S. Sra, Positive definite matrices and the S-Divergence, arXiv:1110.1773 (2011). · Zbl 1338.15046
[23] [srapd] S. Sra, A new metric on the manifold of kernel matrices with application to matrix geometric means, Adv. Neural Inf. Proc. Syst., December 2012. \endbiblist
This reference list is based on information provided by the publisher or from digital mathematics libraries. Its items are heuristically matched to zbMATH identifiers and may contain data conversion errors. In some cases that data have been complemented/enhanced by data from zbMATH Open. This attempts to reflect the references listed in the original paper as accurately as possible without claiming completeness or a perfect matching.