A is further called positive definite, symbolized A > 0, if the strict inequality in 1.1 holds for all nonzero x ∈ C. An equivalent condition forA ∈ Mn to be positive definite is thatA is Hermitian and all eigenvalues of A are positive real numbers. Given a positive semidefinite matrix A and p > 0, A denotes the unique positive semidefinite pth power of A. Let A and B be two Hermitian matrices ...

1 Semidefinite programming Let Sn×n be the set of n by n real symmetric matrices. Definition 1 A ∈ Sn×n is called positive semidefinite, denoted A 0, if xAx ≥ 0 for any x ∈ R. There are several well-known equivalent ways to state positive semidefiniteness. Proposition 1 The following are equivalent: (i) A is positive semidefinite. (ii) Every eigenvalue of A is nonnegative. (iii) There is a matr...

We investigate the completely positive semidefinite cone CS+, a new matrix cone consisting of all n×n matrices that admit a Gram representation by positive semidefinite matrices (of any size). In particular, we study relationships between this cone and the completely positive and the doubly nonnegative cone, and between its dual cone and trace positive non-commutative polynomials. We use this n...

The learning of appropriate distance metrics is a critical problem in image classification and retrieval. In this work, we propose a boosting-based technique, termed BOOSTMETRIC, for learning a Mahalanobis distance metric. One of the primary difficulties in learning such a metric is to ensure that the Mahalanobis matrix remains positive semidefinite. Semidefinite programming is sometimes used t...

We build upon the work of Fukuda et al. [SIAM J. Optim., 11 (2001), pp. 647–674] and Nakata et al. [Math. Program., 95 (2003), pp. 303–327], in which the theory of partial positive semidefinite matrices was applied to the semidefinite programming (SDP) problem as a technique for exploiting sparsity in the data. In contrast to their work, which improved an existing algorithm based on a standard ...