1DLASD4(1)           LAPACK auxiliary routine (version 3.2)           DLASD4(1)
2
3
4

NAME

6       DLASD4  - subroutine compute the square root of the I-th updated eigen‐
7       value of a positive symmetric rank-one modification to a positive diag‐
8       onal matrix whose entries are given as the squares of the corresponding
9       entries in the array d, and that   0 <= D(i) < D(j) for i < j  and that
10       RHO > 0
11

SYNOPSIS

13       SUBROUTINE DLASD4( N, I, D, Z, DELTA, RHO, SIGMA, WORK, INFO )
14
15           INTEGER        I, INFO, N
16
17           DOUBLE         PRECISION RHO, SIGMA
18
19           DOUBLE         PRECISION D( * ), DELTA( * ), WORK( * ), Z( * )
20

PURPOSE

22       This subroutine computes the square root of the I-th updated eigenvalue
23       of a positive symmetric rank-one modification to  a  positive  diagonal
24       matrix  whose  entries  are  given  as the squares of the corresponding
25       entries in the array d, and that no loss in generality.   The  rank-one
26       modified system is thus
27              diag( D ) * diag( D ) +  RHO *  Z * Z_transpose.
28       where we assume the Euclidean norm of Z is 1.
29       The method consists of approximating the rational functions in the sec‐
30       ular equation by simpler interpolating rational functions.
31

ARGUMENTS

33       N      (input) INTEGER
34              The length of all arrays.
35
36       I      (input) INTEGER
37              The index of the eigenvalue to be computed.  1 <= I <= N.
38
39       D      (input) DOUBLE PRECISION array, dimension ( N )
40              The original eigenvalues.  It is assumed that they are in order,
41              0 <= D(I) < D(J)  for I < J.
42
43       Z      (input) DOUBLE PRECISION array, dimension ( N )
44              The components of the updating vector.
45
46       DELTA  (output) DOUBLE PRECISION array, dimension ( N )
47              If N .ne. 1, DELTA contains (D(j) - sigma_I) in its  j-th compo‐
48              nent.  If N = 1, then DELTA(1) = 1.  The vector  DELTA  contains
49              the  information necessary to construct the (singular) eigenvec‐
50              tors.
51
52       RHO    (input) DOUBLE PRECISION
53              The scalar in the symmetric updating formula.
54
55       SIGMA  (output) DOUBLE PRECISION
56              The computed sigma_I, the I-th updated eigenvalue.
57
58       WORK   (workspace) DOUBLE PRECISION array, dimension ( N )
59              If N .ne. 1, WORK contains (D(j) + sigma_I) in its  j-th  compo‐
60              nent.  If N = 1, then WORK( 1 ) = 1.
61
62       INFO   (output) INTEGER
63              = 0:  successful exit
64              > 0:  if INFO = 1, the updating process failed.
65

PARAMETERS

67       Logical  variable  ORGATI  (origin-at-i?)  is  used  for distinguishing
68       whether D(i) or D(i+1) is treated  as  the  origin.   ORGATI  =  .true.
69       origin  at  i  ORGATI = .false.   origin at i+1 Logical variable SWTCH3
70       (switch-for-3-poles?) is for noting if we are working with THREE poles!
71       MAXIT  is the maximum number of iterations allowed for each eigenvalue.
72       Further Details =============== Based on contributions by Ren-Cang  Li,
73       Computer Science Division, University of California at Berkeley, USA
74
75
76
77 LAPACK auxiliary routine (versionNo3v.e2m)ber 2008                       DLASD4(1)