1
       2
       3
       4
       5
       6
       7
       8
       9
      10
      11
      12
      13
      14
      15
      16
      17
      18
      19
      20
      21
      22
      23
      24
      25
      26
      27
      28
      29
      30
      31
      32
      33
      34
      35
      36
      37
      38
      39
      40
      41
      42
      43
      44
      45
      46
      47
      48
      49
      50
      51
      52
      53
      54
      55
      56
      57
      58
      59
      60
      61
      62
      63
      64
      65
      66
      67
      68
      69
      70
      71
      72
      73
      74
      75
      76
      77
      78
      79
      80
      81
      82
      83
      84
      85
      86
      87
      88
      89
      90
      91
      92
      93
      94
      95
      96
      97
      98
      99
     100
     101
     102
     103
     104
     105
     106
     107
     108
     109
     110
     111
     112
     113
     114
     115
     116
     117
     118
     119
     120
     121
     122
     123
     124
     125
      SUBROUTINE DLAED5( I, D, Z, DELTA, RHO, DLAM )
*
*  -- LAPACK routine (version 3.2) --
*  -- LAPACK is a software package provided by Univ. of Tennessee,    --
*  -- Univ. of California Berkeley, Univ. of Colorado Denver and NAG Ltd..--
*     November 2006
*
*     .. Scalar Arguments ..
      INTEGER            I
      DOUBLE PRECISION   DLAM, RHO
*     ..
*     .. Array Arguments ..
      DOUBLE PRECISION   D( 2 ), DELTA( 2 ), Z( 2 )
*     ..
*
*  Purpose
*  =======
*
*  This subroutine computes the I-th eigenvalue of a symmetric rank-one
*  modification of a 2-by-2 diagonal matrix
*
*             diag( D )  +  RHO *  Z * transpose(Z) .
*
*  The diagonal elements in the array D are assumed to satisfy
*
*             D(i) < D(j)  for  i < j .
*
*  We also assume RHO > 0 and that the Euclidean norm of the vector
*  Z is one.
*
*  Arguments
*  =========
*
*  I      (input) INTEGER
*         The index of the eigenvalue to be computed.  I = 1 or I = 2.
*
*  D      (input) DOUBLE PRECISION array, dimension (2)
*         The original eigenvalues.  We assume D(1) < D(2).
*
*  Z      (input) DOUBLE PRECISION array, dimension (2)
*         The components of the updating vector.
*
*  DELTA  (output) DOUBLE PRECISION array, dimension (2)
*         The vector DELTA contains the information necessary
*         to construct the eigenvectors.
*
*  RHO    (input) DOUBLE PRECISION
*         The scalar in the symmetric updating formula.
*
*  DLAM   (output) DOUBLE PRECISION
*         The computed lambda_I, the I-th updated eigenvalue.
*
*  Further Details
*  ===============
*
*  Based on contributions by
*     Ren-Cang Li, Computer Science Division, University of California
*     at Berkeley, USA
*
*  =====================================================================
*
*     .. Parameters ..
      DOUBLE PRECISION   ZERO, ONE, TWO, FOUR
      PARAMETER          ( ZERO = 0.0D0, ONE = 1.0D0, TWO = 2.0D0,
     $                   FOUR = 4.0D0 )
*     ..
*     .. Local Scalars ..
      DOUBLE PRECISION   B, C, DEL, TAU, TEMP, W
*     ..
*     .. Intrinsic Functions ..
      INTRINSIC          ABSSQRT
*     ..
*     .. Executable Statements ..
*
      DEL = D( 2 ) - D( 1 )
      IF( I.EQ.1 ) THEN
         W = ONE + TWO*RHO*( Z( 2 )*Z( 2 )-Z( 1 )*Z( 1 ) ) / DEL
         IF( W.GT.ZERO ) THEN
            B = DEL + RHO*( Z( 1 )*Z( 1 )+Z( 2 )*Z( 2 ) )
            C = RHO*Z( 1 )*Z( 1 )*DEL
*
*           B > ZERO, always
*
            TAU = TWO*/ ( B+SQRTABS( B*B-FOUR*C ) ) )
            DLAM = D( 1 ) + TAU
            DELTA( 1 ) = -Z( 1 ) / TAU
            DELTA( 2 ) = Z( 2 ) / ( DEL-TAU )
         ELSE
            B = -DEL + RHO*( Z( 1 )*Z( 1 )+Z( 2 )*Z( 2 ) )
            C = RHO*Z( 2 )*Z( 2 )*DEL
            IF( B.GT.ZERO ) THEN
               TAU = -TWO*/ ( B+SQRT( B*B+FOUR*C ) )
            ELSE
               TAU = ( B-SQRT( B*B+FOUR*C ) ) / TWO
            END IF
            DLAM = D( 2 ) + TAU
            DELTA( 1 ) = -Z( 1 ) / ( DEL+TAU )
            DELTA( 2 ) = -Z( 2 ) / TAU
         END IF
         TEMP = SQRT( DELTA( 1 )*DELTA( 1 )+DELTA( 2 )*DELTA( 2 ) )
         DELTA( 1 ) = DELTA( 1 ) / TEMP
         DELTA( 2 ) = DELTA( 2 ) / TEMP
      ELSE
*
*     Now I=2
*
         B = -DEL + RHO*( Z( 1 )*Z( 1 )+Z( 2 )*Z( 2 ) )
         C = RHO*Z( 2 )*Z( 2 )*DEL
         IF( B.GT.ZERO ) THEN
            TAU = ( B+SQRT( B*B+FOUR*C ) ) / TWO
         ELSE
            TAU = TWO*/ ( -B+SQRT( B*B+FOUR*C ) )
         END IF
         DLAM = D( 2 ) + TAU
         DELTA( 1 ) = -Z( 1 ) / ( DEL+TAU )
         DELTA( 2 ) = -Z( 2 ) / TAU
         TEMP = SQRT( DELTA( 1 )*DELTA( 1 )+DELTA( 2 )*DELTA( 2 ) )
         DELTA( 1 ) = DELTA( 1 ) / TEMP
         DELTA( 2 ) = DELTA( 2 ) / TEMP
      END IF
      RETURN
*
*     End OF DLAED5
*
      END