Annotation of rpl/lapack/lapack/dgghrd.f, revision 1.5

1.1       bertrand    1:       SUBROUTINE DGGHRD( COMPQ, COMPZ, N, ILO, IHI, A, LDA, B, LDB, Q,
                      2:      $                   LDQ, Z, LDZ, INFO )
                      3: *
                      4: *  -- LAPACK routine (version 3.2) --
                      5: *  -- LAPACK is a software package provided by Univ. of Tennessee,    --
                      6: *  -- Univ. of California Berkeley, Univ. of Colorado Denver and NAG Ltd..--
                      7: *     November 2006
                      8: *
                      9: *     .. Scalar Arguments ..
                     10:       CHARACTER          COMPQ, COMPZ
                     11:       INTEGER            IHI, ILO, INFO, LDA, LDB, LDQ, LDZ, N
                     12: *     ..
                     13: *     .. Array Arguments ..
                     14:       DOUBLE PRECISION   A( LDA, * ), B( LDB, * ), Q( LDQ, * ),
                     15:      $                   Z( LDZ, * )
                     16: *     ..
                     17: *
                     18: *  Purpose
                     19: *  =======
                     20: *
                     21: *  DGGHRD reduces a pair of real matrices (A,B) to generalized upper
                     22: *  Hessenberg form using orthogonal transformations, where A is a
                     23: *  general matrix and B is upper triangular.  The form of the
                     24: *  generalized eigenvalue problem is
                     25: *     A*x = lambda*B*x,
                     26: *  and B is typically made upper triangular by computing its QR
                     27: *  factorization and moving the orthogonal matrix Q to the left side
                     28: *  of the equation.
                     29: *
                     30: *  This subroutine simultaneously reduces A to a Hessenberg matrix H:
                     31: *     Q**T*A*Z = H
                     32: *  and transforms B to another upper triangular matrix T:
                     33: *     Q**T*B*Z = T
                     34: *  in order to reduce the problem to its standard form
                     35: *     H*y = lambda*T*y
                     36: *  where y = Z**T*x.
                     37: *
                     38: *  The orthogonal matrices Q and Z are determined as products of Givens
                     39: *  rotations.  They may either be formed explicitly, or they may be
                     40: *  postmultiplied into input matrices Q1 and Z1, so that
                     41: *
                     42: *       Q1 * A * Z1**T = (Q1*Q) * H * (Z1*Z)**T
                     43: *
                     44: *       Q1 * B * Z1**T = (Q1*Q) * T * (Z1*Z)**T
                     45: *
                     46: *  If Q1 is the orthogonal matrix from the QR factorization of B in the
                     47: *  original equation A*x = lambda*B*x, then DGGHRD reduces the original
                     48: *  problem to generalized Hessenberg form.
                     49: *
                     50: *  Arguments
                     51: *  =========
                     52: *
                     53: *  COMPQ   (input) CHARACTER*1
                     54: *          = 'N': do not compute Q;
                     55: *          = 'I': Q is initialized to the unit matrix, and the
                     56: *                 orthogonal matrix Q is returned;
                     57: *          = 'V': Q must contain an orthogonal matrix Q1 on entry,
                     58: *                 and the product Q1*Q is returned.
                     59: *
                     60: *  COMPZ   (input) CHARACTER*1
                     61: *          = 'N': do not compute Z;
                     62: *          = 'I': Z is initialized to the unit matrix, and the
                     63: *                 orthogonal matrix Z is returned;
                     64: *          = 'V': Z must contain an orthogonal matrix Z1 on entry,
                     65: *                 and the product Z1*Z is returned.
                     66: *
                     67: *  N       (input) INTEGER
                     68: *          The order of the matrices A and B.  N >= 0.
                     69: *
                     70: *  ILO     (input) INTEGER
                     71: *  IHI     (input) INTEGER
                     72: *          ILO and IHI mark the rows and columns of A which are to be
                     73: *          reduced.  It is assumed that A is already upper triangular
                     74: *          in rows and columns 1:ILO-1 and IHI+1:N.  ILO and IHI are
                     75: *          normally set by a previous call to SGGBAL; otherwise they
                     76: *          should be set to 1 and N respectively.
                     77: *          1 <= ILO <= IHI <= N, if N > 0; ILO=1 and IHI=0, if N=0.
                     78: *
                     79: *  A       (input/output) DOUBLE PRECISION array, dimension (LDA, N)
                     80: *          On entry, the N-by-N general matrix to be reduced.
                     81: *          On exit, the upper triangle and the first subdiagonal of A
                     82: *          are overwritten with the upper Hessenberg matrix H, and the
                     83: *          rest is set to zero.
                     84: *
                     85: *  LDA     (input) INTEGER
                     86: *          The leading dimension of the array A.  LDA >= max(1,N).
                     87: *
                     88: *  B       (input/output) DOUBLE PRECISION array, dimension (LDB, N)
                     89: *          On entry, the N-by-N upper triangular matrix B.
                     90: *          On exit, the upper triangular matrix T = Q**T B Z.  The
                     91: *          elements below the diagonal are set to zero.
                     92: *
                     93: *  LDB     (input) INTEGER
                     94: *          The leading dimension of the array B.  LDB >= max(1,N).
                     95: *
                     96: *  Q       (input/output) DOUBLE PRECISION array, dimension (LDQ, N)
                     97: *          On entry, if COMPQ = 'V', the orthogonal matrix Q1,
                     98: *          typically from the QR factorization of B.
                     99: *          On exit, if COMPQ='I', the orthogonal matrix Q, and if
                    100: *          COMPQ = 'V', the product Q1*Q.
                    101: *          Not referenced if COMPQ='N'.
                    102: *
                    103: *  LDQ     (input) INTEGER
                    104: *          The leading dimension of the array Q.
                    105: *          LDQ >= N if COMPQ='V' or 'I'; LDQ >= 1 otherwise.
                    106: *
                    107: *  Z       (input/output) DOUBLE PRECISION array, dimension (LDZ, N)
                    108: *          On entry, if COMPZ = 'V', the orthogonal matrix Z1.
                    109: *          On exit, if COMPZ='I', the orthogonal matrix Z, and if
                    110: *          COMPZ = 'V', the product Z1*Z.
                    111: *          Not referenced if COMPZ='N'.
                    112: *
                    113: *  LDZ     (input) INTEGER
                    114: *          The leading dimension of the array Z.
                    115: *          LDZ >= N if COMPZ='V' or 'I'; LDZ >= 1 otherwise.
                    116: *
                    117: *  INFO    (output) INTEGER
                    118: *          = 0:  successful exit.
                    119: *          < 0:  if INFO = -i, the i-th argument had an illegal value.
                    120: *
                    121: *  Further Details
                    122: *  ===============
                    123: *
                    124: *  This routine reduces A to Hessenberg and B to triangular form by
                    125: *  an unblocked reduction, as described in _Matrix_Computations_,
                    126: *  by Golub and Van Loan (Johns Hopkins Press.)
                    127: *
                    128: *  =====================================================================
                    129: *
                    130: *     .. Parameters ..
                    131:       DOUBLE PRECISION   ONE, ZERO
                    132:       PARAMETER          ( ONE = 1.0D+0, ZERO = 0.0D+0 )
                    133: *     ..
                    134: *     .. Local Scalars ..
                    135:       LOGICAL            ILQ, ILZ
                    136:       INTEGER            ICOMPQ, ICOMPZ, JCOL, JROW
                    137:       DOUBLE PRECISION   C, S, TEMP
                    138: *     ..
                    139: *     .. External Functions ..
                    140:       LOGICAL            LSAME
                    141:       EXTERNAL           LSAME
                    142: *     ..
                    143: *     .. External Subroutines ..
                    144:       EXTERNAL           DLARTG, DLASET, DROT, XERBLA
                    145: *     ..
                    146: *     .. Intrinsic Functions ..
                    147:       INTRINSIC          MAX
                    148: *     ..
                    149: *     .. Executable Statements ..
                    150: *
                    151: *     Decode COMPQ
                    152: *
                    153:       IF( LSAME( COMPQ, 'N' ) ) THEN
                    154:          ILQ = .FALSE.
                    155:          ICOMPQ = 1
                    156:       ELSE IF( LSAME( COMPQ, 'V' ) ) THEN
                    157:          ILQ = .TRUE.
                    158:          ICOMPQ = 2
                    159:       ELSE IF( LSAME( COMPQ, 'I' ) ) THEN
                    160:          ILQ = .TRUE.
                    161:          ICOMPQ = 3
                    162:       ELSE
                    163:          ICOMPQ = 0
                    164:       END IF
                    165: *
                    166: *     Decode COMPZ
                    167: *
                    168:       IF( LSAME( COMPZ, 'N' ) ) THEN
                    169:          ILZ = .FALSE.
                    170:          ICOMPZ = 1
                    171:       ELSE IF( LSAME( COMPZ, 'V' ) ) THEN
                    172:          ILZ = .TRUE.
                    173:          ICOMPZ = 2
                    174:       ELSE IF( LSAME( COMPZ, 'I' ) ) THEN
                    175:          ILZ = .TRUE.
                    176:          ICOMPZ = 3
                    177:       ELSE
                    178:          ICOMPZ = 0
                    179:       END IF
                    180: *
                    181: *     Test the input parameters.
                    182: *
                    183:       INFO = 0
                    184:       IF( ICOMPQ.LE.0 ) THEN
                    185:          INFO = -1
                    186:       ELSE IF( ICOMPZ.LE.0 ) THEN
                    187:          INFO = -2
                    188:       ELSE IF( N.LT.0 ) THEN
                    189:          INFO = -3
                    190:       ELSE IF( ILO.LT.1 ) THEN
                    191:          INFO = -4
                    192:       ELSE IF( IHI.GT.N .OR. IHI.LT.ILO-1 ) THEN
                    193:          INFO = -5
                    194:       ELSE IF( LDA.LT.MAX( 1, N ) ) THEN
                    195:          INFO = -7
                    196:       ELSE IF( LDB.LT.MAX( 1, N ) ) THEN
                    197:          INFO = -9
                    198:       ELSE IF( ( ILQ .AND. LDQ.LT.N ) .OR. LDQ.LT.1 ) THEN
                    199:          INFO = -11
                    200:       ELSE IF( ( ILZ .AND. LDZ.LT.N ) .OR. LDZ.LT.1 ) THEN
                    201:          INFO = -13
                    202:       END IF
                    203:       IF( INFO.NE.0 ) THEN
                    204:          CALL XERBLA( 'DGGHRD', -INFO )
                    205:          RETURN
                    206:       END IF
                    207: *
                    208: *     Initialize Q and Z if desired.
                    209: *
                    210:       IF( ICOMPQ.EQ.3 )
                    211:      $   CALL DLASET( 'Full', N, N, ZERO, ONE, Q, LDQ )
                    212:       IF( ICOMPZ.EQ.3 )
                    213:      $   CALL DLASET( 'Full', N, N, ZERO, ONE, Z, LDZ )
                    214: *
                    215: *     Quick return if possible
                    216: *
                    217:       IF( N.LE.1 )
                    218:      $   RETURN
                    219: *
                    220: *     Zero out lower triangle of B
                    221: *
                    222:       DO 20 JCOL = 1, N - 1
                    223:          DO 10 JROW = JCOL + 1, N
                    224:             B( JROW, JCOL ) = ZERO
                    225:    10    CONTINUE
                    226:    20 CONTINUE
                    227: *
                    228: *     Reduce A and B
                    229: *
                    230:       DO 40 JCOL = ILO, IHI - 2
                    231: *
                    232:          DO 30 JROW = IHI, JCOL + 2, -1
                    233: *
                    234: *           Step 1: rotate rows JROW-1, JROW to kill A(JROW,JCOL)
                    235: *
                    236:             TEMP = A( JROW-1, JCOL )
                    237:             CALL DLARTG( TEMP, A( JROW, JCOL ), C, S,
                    238:      $                   A( JROW-1, JCOL ) )
                    239:             A( JROW, JCOL ) = ZERO
                    240:             CALL DROT( N-JCOL, A( JROW-1, JCOL+1 ), LDA,
                    241:      $                 A( JROW, JCOL+1 ), LDA, C, S )
                    242:             CALL DROT( N+2-JROW, B( JROW-1, JROW-1 ), LDB,
                    243:      $                 B( JROW, JROW-1 ), LDB, C, S )
                    244:             IF( ILQ )
                    245:      $         CALL DROT( N, Q( 1, JROW-1 ), 1, Q( 1, JROW ), 1, C, S )
                    246: *
                    247: *           Step 2: rotate columns JROW, JROW-1 to kill B(JROW,JROW-1)
                    248: *
                    249:             TEMP = B( JROW, JROW )
                    250:             CALL DLARTG( TEMP, B( JROW, JROW-1 ), C, S,
                    251:      $                   B( JROW, JROW ) )
                    252:             B( JROW, JROW-1 ) = ZERO
                    253:             CALL DROT( IHI, A( 1, JROW ), 1, A( 1, JROW-1 ), 1, C, S )
                    254:             CALL DROT( JROW-1, B( 1, JROW ), 1, B( 1, JROW-1 ), 1, C,
                    255:      $                 S )
                    256:             IF( ILZ )
                    257:      $         CALL DROT( N, Z( 1, JROW ), 1, Z( 1, JROW-1 ), 1, C, S )
                    258:    30    CONTINUE
                    259:    40 CONTINUE
                    260: *
                    261:       RETURN
                    262: *
                    263: *     End of DGGHRD
                    264: *
                    265:       END

CVSweb interface <joel.bertrand@systella.fr>