1: \begin{abstract}

2:   In this paper a two-sided, parallel Kogbetliantz-type algorithm for

3:   the hyperbolic singular value decomposition (HSVD) of real and

4:   complex square matrices is developed, with a single assumption that

5:   the input matrix, of order $n$, admits such a decomposition into the

6:   product of a unitary, a non-negative diagonal, and a $J$-unitary

7:   matrix, where $J$ is a given diagonal matrix of positive and

8:   negative signs.  When $J=\pm I$, the proposed algorithm computes the

9:   ordinary SVD\@.

10: 

11:   \looseness=-1

12:   The paper's most important contribution---a derivation of formulas

13:   for the HSVD of $2\times 2$ matrices---is presented first, followed

14:   by the details of their implementation in floating-point arithmetic.

15:   Next, the effects of the hyperbolic transformations on the columns

16:   of the iteration matrix are discussed.  These effects then guide a

17:   redesign of the dynamic pivot ordering, being already a

18:   well-established pivot strategy for the ordinary Kogbetliantz

19:   algorithm, for the general, $n\times n$ HSVD\@.  A heuristic but

20:   sound convergence criterion is then proposed, which contributes to

21:   high accuracy demonstrated in the numerical testing results.  Such a

22:   $J$-Kogbetliantz algorithm as presented here is intrinsically slow,

23:   but is nevertheless usable for matrices of small orders.

24:   \keywords{hyperbolic singular value decomposition\and Kogbetliantz algorithm\and Hermitian eigenproblem\and OpenMP multicore parallelization}

25:   \subclass{65F15\and 65Y05\and 15A18}

26: \end{abstract}

27: