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
     126
     127
     128
     129
     130
     131
     132
     133
     134
     135
     136
     137
     138
     139
     140
     141
     142
     143
     144
     145
     146
     147
     148
     149
     150
     151
     152
     153
     154
     155
     156
     157
     158
     159
     160
     161
     162
     163
     164
     165
     166
     167
     168
     169
     170
     171
     172
     173
     174
     175
     176
     177
     178
     179
     180
     181
     182
     183
     184
     185
     186
     187
     188
     189
     190
     191
     192
     193
     194
     195
     196
     197
     198
     199
     200
     201
     202
     203
     204
     205
     206
     207
     208
     209
     210
     211
     212
     213
     214
     215
     216
     217
     218
     219
     220
     221
     222
     223
     224
     225
     226
     227
     228
     229
     230
     231
     232
     233
     234
     235
     236
     237
     238
     239
     240
     241
     242
     243
     244
     245
     246
     247
     248
     249
     250
     251
     252
     253
     254
     255
     256
     257
     258
     259
     260
     261
     262
     263
     264
     265
     266
     267
     268
     269
     270
     271
     272
     273
     274
     275
      SUBROUTINE DGEHRD( N, ILO, IHI, A, LDA, TAU, WORK, LWORK, INFO )
*
*  -- LAPACK routine (version 3.3.1)                                  --
*  -- LAPACK is a software package provided by Univ. of Tennessee,    --
*  -- Univ. of California Berkeley, Univ. of Colorado Denver and NAG Ltd..--
*  -- April 2009                                                      --
*
*     .. Scalar Arguments ..
      INTEGER            IHI, ILO, INFO, LDA, LWORK, N
*     ..
*     .. Array Arguments ..
      DOUBLE PRECISION  A( LDA, * ), TAU( * ), WORK( * )
*     ..
*
*  Purpose
*  =======
*
*  DGEHRD reduces a real general matrix A to upper Hessenberg form H by
*  an orthogonal similarity transformation:  Q**T * A * Q = H .
*
*  Arguments
*  =========
*
*  N       (input) INTEGER
*          The order of the matrix A.  N >= 0.
*
*  ILO     (input) INTEGER
*  IHI     (input) INTEGER
*          It is assumed that A is already upper triangular in rows
*          and columns 1:ILO-1 and IHI+1:N. ILO and IHI are normally
*          set by a previous call to DGEBAL; otherwise they should be
*          set to 1 and N respectively. See Further Details.
*          1 <= ILO <= IHI <= N, if N > 0; ILO=1 and IHI=0, if N=0.
*
*  A       (input/output) DOUBLE PRECISION array, dimension (LDA,N)
*          On entry, the N-by-N general matrix to be reduced.
*          On exit, the upper triangle and the first subdiagonal of A
*          are overwritten with the upper Hessenberg matrix H, and the
*          elements below the first subdiagonal, with the array TAU,
*          represent the orthogonal matrix Q as a product of elementary
*          reflectors. See Further Details.
*
*  LDA     (input) INTEGER
*          The leading dimension of the array A.  LDA >= max(1,N).
*
*  TAU     (output) DOUBLE PRECISION array, dimension (N-1)
*          The scalar factors of the elementary reflectors (see Further
*          Details). Elements 1:ILO-1 and IHI:N-1 of TAU are set to
*          zero.
*
*  WORK    (workspace/output) DOUBLE PRECISION array, dimension (LWORK)
*          On exit, if INFO = 0, WORK(1) returns the optimal LWORK.
*
*  LWORK   (input) INTEGER
*          The length of the array WORK.  LWORK >= max(1,N).
*          For optimum performance LWORK >= N*NB, where NB is the
*          optimal blocksize.
*
*          If LWORK = -1, then a workspace query is assumed; the routine
*          only calculates the optimal size of the WORK array, returns
*          this value as the first entry of the WORK array, and no error
*          message related to LWORK is issued by XERBLA.
*
*  INFO    (output) INTEGER
*          = 0:  successful exit
*          < 0:  if INFO = -i, the i-th argument had an illegal value.
*
*  Further Details
*  ===============
*
*  The matrix Q is represented as a product of (ihi-ilo) elementary
*  reflectors
*
*     Q = H(ilo) H(ilo+1) . . . H(ihi-1).
*
*  Each H(i) has the form
*
*     H(i) = I - tau * v * v**T
*
*  where tau is a real scalar, and v is a real vector with
*  v(1:i) = 0, v(i+1) = 1 and v(ihi+1:n) = 0; v(i+2:ihi) is stored on
*  exit in A(i+2:ihi,i), and tau in TAU(i).
*
*  The contents of A are illustrated by the following example, with
*  n = 7, ilo = 2 and ihi = 6:
*
*  on entry,                        on exit,
*
*  ( a   a   a   a   a   a   a )    (  a   a   h   h   h   h   a )
*  (     a   a   a   a   a   a )    (      a   h   h   h   h   a )
*  (     a   a   a   a   a   a )    (      h   h   h   h   h   h )
*  (     a   a   a   a   a   a )    (      v2  h   h   h   h   h )
*  (     a   a   a   a   a   a )    (      v2  v3  h   h   h   h )
*  (     a   a   a   a   a   a )    (      v2  v3  v4  h   h   h )
*  (                         a )    (                          a )
*
*  where a denotes an element of the original matrix A, h denotes a
*  modified element of the upper Hessenberg matrix H, and vi denotes an
*  element of the vector defining H(i).
*
*  This file is a slight modification of LAPACK-3.0's DGEHRD
*  subroutine incorporating improvements proposed by Quintana-Orti and
*  Van de Geijn (2006). (See DLAHR2.)
*
*  =====================================================================
*
*     .. Parameters ..
      INTEGER            NBMAX, LDT
      PARAMETER          ( NBMAX = 64, LDT = NBMAX+1 )
      DOUBLE PRECISION  ZERO, ONE
      PARAMETER          ( ZERO = 0.0D+0
     $                     ONE = 1.0D+0 )
*     ..
*     .. Local Scalars ..
      LOGICAL            LQUERY
      INTEGER            I, IB, IINFO, IWS, J, LDWORK, LWKOPT, NB,
     $                   NBMIN, NH, NX
      DOUBLE PRECISION  EI
*     ..
*     .. Local Arrays ..
      DOUBLE PRECISION  T( LDT, NBMAX )
*     ..
*     .. External Subroutines ..
      EXTERNAL           DAXPY, DGEHD2, DGEMM, DLAHR2, DLARFB, DTRMM,
     $                   XERBLA
*     ..
*     .. Intrinsic Functions ..
      INTRINSIC          MAXMIN
*     ..
*     .. External Functions ..
      INTEGER            ILAENV
      EXTERNAL           ILAENV
*     ..
*     .. Executable Statements ..
*
*     Test the input parameters
*
      INFO = 0
      NB = MIN( NBMAX, ILAENV( 1'DGEHRD'' ', N, ILO, IHI, -1 ) )
      LWKOPT = N*NB
      WORK( 1 ) = LWKOPT
      LQUERY = ( LWORK.EQ.-1 )
      IF( N.LT.0 ) THEN
         INFO = -1
      ELSE IF( ILO.LT.1 .OR. ILO.GT.MAX1, N ) ) THEN
         INFO = -2
      ELSE IF( IHI.LT.MIN( ILO, N ) .OR. IHI.GT.N ) THEN
         INFO = -3
      ELSE IF( LDA.LT.MAX1, N ) ) THEN
         INFO = -5
      ELSE IF( LWORK.LT.MAX1, N ) .AND. .NOT.LQUERY ) THEN
         INFO = -8
      END IF
      IF( INFO.NE.0 ) THEN
         CALL XERBLA( 'DGEHRD'-INFO )
         RETURN
      ELSE IF( LQUERY ) THEN
         RETURN
      END IF
*
*     Set elements 1:ILO-1 and IHI:N-1 of TAU to zero
*
      DO 10 I = 1, ILO - 1
         TAU( I ) = ZERO
   10 CONTINUE
      DO 20 I = MAX1, IHI ), N - 1
         TAU( I ) = ZERO
   20 CONTINUE
*
*     Quick return if possible
*
      NH = IHI - ILO + 1
      IF( NH.LE.1 ) THEN
         WORK( 1 ) = 1
         RETURN
      END IF
*
*     Determine the block size
*
      NB = MIN( NBMAX, ILAENV( 1'DGEHRD'' ', N, ILO, IHI, -1 ) )
      NBMIN = 2
      IWS = 1
      IF( NB.GT.1 .AND. NB.LT.NH ) THEN
*
*        Determine when to cross over from blocked to unblocked code
*        (last block is always handled by unblocked code)
*
         NX = MAX( NB, ILAENV( 3'DGEHRD'' ', N, ILO, IHI, -1 ) )
         IF( NX.LT.NH ) THEN
*
*           Determine if workspace is large enough for blocked code
*
            IWS = N*NB
            IF( LWORK.LT.IWS ) THEN
*
*              Not enough workspace to use optimal NB:  determine the
*              minimum value of NB, and reduce NB or force use of
*              unblocked code
*
               NBMIN = MAX2, ILAENV( 2'DGEHRD'' ', N, ILO, IHI,
     $                 -1 ) )
               IF( LWORK.GE.N*NBMIN ) THEN
                  NB = LWORK / N
               ELSE
                  NB = 1
               END IF
            END IF
         END IF
      END IF
      LDWORK = N
*

      IF( NB.LT.NBMIN .OR. NB.GE.NH ) THEN
*
*        Use unblocked code below
*
         I = ILO
*
      ELSE
*
*        Use blocked code
*
         DO 40 I = ILO, IHI - 1 - NX, NB
            IB = MIN( NB, IHI-I )
*
*           Reduce columns i:i+ib-1 to Hessenberg form, returning the
*           matrices V and T of the block reflector H = I - V*T*V**T
*           which performs the reduction, and also the matrix Y = A*V*T
*
            CALL DLAHR2( IHI, I, IB, A( 1, I ), LDA, TAU( I ), T, LDT,
     $                   WORK, LDWORK )
*
*           Apply the block reflector H to A(1:ihi,i+ib:ihi) from the
*           right, computing  A := A - Y * V**T. V(i+ib,ib-1) must be set
*           to 1
*
            EI = A( I+IB, I+IB-1 )
            A( I+IB, I+IB-1 ) = ONE
            CALL DGEMM( 'No transpose''Transpose'
     $                  IHI, IHI-I-IB+1,
     $                  IB, -ONE, WORK, LDWORK, A( I+IB, I ), LDA, ONE,
     $                  A( 1, I+IB ), LDA )
            A( I+IB, I+IB-1 ) = EI
*
*           Apply the block reflector H to A(1:i,i+1:i+ib-1) from the
*           right
*
            CALL DTRMM( 'Right''Lower''Transpose',
     $                  'Unit', I, IB-1,
     $                  ONE, A( I+1, I ), LDA, WORK, LDWORK )
            DO 30 J = 0, IB-2
               CALL DAXPY( I, -ONE, WORK( LDWORK*J+1 ), 1,
     $                     A( 1, I+J+1 ), 1 )
   30       CONTINUE
*
*           Apply the block reflector H to A(i+1:ihi,i+ib:n) from the
*           left
*
            CALL DLARFB( 'Left''Transpose''Forward',
     $                   'Columnwise',
     $                   IHI-I, N-I-IB+1, IB, A( I+1, I ), LDA, T, LDT,
     $                   A( I+1, I+IB ), LDA, WORK, LDWORK )
   40    CONTINUE
      END IF
*
*     Use unblocked code to reduce the rest of the matrix
*
      CALL DGEHD2( N, I, IHI, A, LDA, TAU, WORK, IINFO )
      WORK( 1 ) = IWS
*
      RETURN
*
*     End of DGEHRD
*
      END