1DLARRJ(1)           LAPACK auxiliary routine (version 3.1)           DLARRJ(1)
2
3
4

NAME

6       DLARRJ  - initial eigenvalue approximations of T, DLARRJ does bisection
7       to refine the eigenvalues of T,
8

SYNOPSIS

10       SUBROUTINE DLARRJ( N, D, E2, IFIRST,  ILAST,  RTOL,  OFFSET,  W,  WERR,
11                          WORK, IWORK, PIVMIN, SPDIAM, INFO )
12
13           INTEGER        IFIRST, ILAST, INFO, N, OFFSET
14
15           DOUBLE         PRECISION PIVMIN, RTOL, SPDIAM
16
17           INTEGER        IWORK( * )
18
19           DOUBLE         PRECISION  D( * ), E2( * ), W( * ), WERR( * ), WORK(
20                          * )
21

PURPOSE

23       Given the initial eigenvalue approximations of T, DLARRJ  does   bisec‐
24       tion  to  refine  the  eigenvalues  of T, W( IFIRST-OFFSET ) through W(
25       ILAST-OFFSET ), to more accuracy. Initial guesses for these eigenvalues
26       are  input  in  W,  the  corresponding  estimate  of the error in these
27       guesses in WERR. During bisection, intervals [left,  right]  are  main‐
28       tained  by storing their mid-points and semi-widths in the arrays W and
29       WERR respectively.
30
31

ARGUMENTS

33       N       (input) INTEGER
34               The order of the matrix.
35
36       D       (input) DOUBLE PRECISION array, dimension (N)
37               The N diagonal elements of T.
38
39       E2      (input) DOUBLE PRECISION array, dimension (N-1)
40               The Squares of the (N-1) subdiagonal elements of T.
41
42       IFIRST  (input) INTEGER
43               The index of the first eigenvalue to be computed.
44
45       ILAST   (input) INTEGER
46               The index of the last eigenvalue to be computed.
47
48       RTOL   (input) DOUBLE PRECISION
49              Tolerance for the convergence of the  bisection  intervals.   An
50              interval     [LEFT,RIGHT]     has     converged     if    RIGHT-
51              LEFT.LT.RTOL*MAX(|LEFT|,|RIGHT|).
52
53       OFFSET  (input) INTEGER
54               Offset for the arrays  W  and  WERR,  i.e.,  the  IFIRST-OFFSET
55               through ILAST-OFFSET elements of these arrays are to be used.
56
57       W       (input/output) DOUBLE PRECISION array, dimension (N)
58               On  input,  W(  IFIRST-OFFSET  )  through W( ILAST-OFFSET ) are
59               estimates of the eigenvalues of L D L^T indexed IFIRST  through
60               ILAST.  On output, these estimates are refined.
61
62       WERR    (input/output) DOUBLE PRECISION array, dimension (N)
63               On  input,  WERR(  IFIRST-OFFSET ) through WERR( ILAST-OFFSET )
64               are the errors in the estimates of the  corresponding  elements
65               in W.  On output, these errors are refined.
66
67       WORK    (workspace) DOUBLE PRECISION array, dimension (2*N)
68               Workspace.
69
70       IWORK   (workspace) INTEGER array, dimension (2*N)
71               Workspace.
72
73       PIVMIN  (input) DOUBLE PRECISION
74               The minimum pivot in the Sturm sequence for T.
75
76       SPDIAM  (input) DOUBLE PRECISION
77               The spectral diameter of T.
78
79       INFO    (output) INTEGER
80               Error flag.
81

FURTHER DETAILS

83       Based on contributions by
84          Beresford Parlett, University of California, Berkeley, USA
85          Jim Demmel, University of California, Berkeley, USA
86          Inderjit Dhillon, University of Texas, Austin, USA
87          Osni Marques, LBNL/NERSC, USA
88          Christof Voemel, University of California, Berkeley, USA
89
90
91
92
93 LAPACK auxiliary routine (versionNo3v.e1m)ber 2006                       DLARRJ(1)