Annotation of rpl/lapack/lapack/zlaqps.f, revision 1.8

1.1       bertrand    1:       SUBROUTINE ZLAQPS( M, N, OFFSET, NB, KB, A, LDA, JPVT, TAU, VN1,
                      2:      $                   VN2, AUXV, F, LDF )
                      3: *
1.5       bertrand    4: *  -- LAPACK auxiliary routine (version 3.2.2) --
1.1       bertrand    5: *  -- LAPACK is a software package provided by Univ. of Tennessee,    --
                      6: *  -- Univ. of California Berkeley, Univ. of Colorado Denver and NAG Ltd..--
1.5       bertrand    7: *     June 2010
1.1       bertrand    8: *
                      9: *     .. Scalar Arguments ..
                     10:       INTEGER            KB, LDA, LDF, M, N, NB, OFFSET
                     11: *     ..
                     12: *     .. Array Arguments ..
                     13:       INTEGER            JPVT( * )
                     14:       DOUBLE PRECISION   VN1( * ), VN2( * )
                     15:       COMPLEX*16         A( LDA, * ), AUXV( * ), F( LDF, * ), TAU( * )
                     16: *     ..
                     17: *
                     18: *  Purpose
                     19: *  =======
                     20: *
                     21: *  ZLAQPS computes a step of QR factorization with column pivoting
                     22: *  of a complex M-by-N matrix A by using Blas-3.  It tries to factorize
                     23: *  NB columns from A starting from the row OFFSET+1, and updates all
                     24: *  of the matrix with Blas-3 xGEMM.
                     25: *
                     26: *  In some cases, due to catastrophic cancellations, it cannot
                     27: *  factorize NB columns.  Hence, the actual number of factorized
                     28: *  columns is returned in KB.
                     29: *
                     30: *  Block A(1:OFFSET,1:N) is accordingly pivoted, but not factorized.
                     31: *
                     32: *  Arguments
                     33: *  =========
                     34: *
                     35: *  M       (input) INTEGER
                     36: *          The number of rows of the matrix A. M >= 0.
                     37: *
                     38: *  N       (input) INTEGER
                     39: *          The number of columns of the matrix A. N >= 0
                     40: *
                     41: *  OFFSET  (input) INTEGER
                     42: *          The number of rows of A that have been factorized in
                     43: *          previous steps.
                     44: *
                     45: *  NB      (input) INTEGER
                     46: *          The number of columns to factorize.
                     47: *
                     48: *  KB      (output) INTEGER
                     49: *          The number of columns actually factorized.
                     50: *
                     51: *  A       (input/output) COMPLEX*16 array, dimension (LDA,N)
                     52: *          On entry, the M-by-N matrix A.
                     53: *          On exit, block A(OFFSET+1:M,1:KB) is the triangular
                     54: *          factor obtained and block A(1:OFFSET,1:N) has been
                     55: *          accordingly pivoted, but no factorized.
                     56: *          The rest of the matrix, block A(OFFSET+1:M,KB+1:N) has
                     57: *          been updated.
                     58: *
                     59: *  LDA     (input) INTEGER
                     60: *          The leading dimension of the array A. LDA >= max(1,M).
                     61: *
                     62: *  JPVT    (input/output) INTEGER array, dimension (N)
                     63: *          JPVT(I) = K <==> Column K of the full matrix A has been
                     64: *          permuted into position I in AP.
                     65: *
                     66: *  TAU     (output) COMPLEX*16 array, dimension (KB)
                     67: *          The scalar factors of the elementary reflectors.
                     68: *
                     69: *  VN1     (input/output) DOUBLE PRECISION array, dimension (N)
                     70: *          The vector with the partial column norms.
                     71: *
                     72: *  VN2     (input/output) DOUBLE PRECISION array, dimension (N)
                     73: *          The vector with the exact column norms.
                     74: *
                     75: *  AUXV    (input/output) COMPLEX*16 array, dimension (NB)
                     76: *          Auxiliar vector.
                     77: *
                     78: *  F       (input/output) COMPLEX*16 array, dimension (LDF,NB)
                     79: *          Matrix F' = L*Y'*A.
                     80: *
                     81: *  LDF     (input) INTEGER
                     82: *          The leading dimension of the array F. LDF >= max(1,N).
                     83: *
                     84: *  Further Details
                     85: *  ===============
                     86: *
                     87: *  Based on contributions by
                     88: *    G. Quintana-Orti, Depto. de Informatica, Universidad Jaime I, Spain
                     89: *    X. Sun, Computer Science Dept., Duke University, USA
                     90: *
                     91: *  =====================================================================
                     92: *
                     93: *     .. Parameters ..
                     94:       DOUBLE PRECISION   ZERO, ONE
                     95:       COMPLEX*16         CZERO, CONE
                     96:       PARAMETER          ( ZERO = 0.0D+0, ONE = 1.0D+0,
                     97:      $                   CZERO = ( 0.0D+0, 0.0D+0 ),
                     98:      $                   CONE = ( 1.0D+0, 0.0D+0 ) )
                     99: *     ..
                    100: *     .. Local Scalars ..
                    101:       INTEGER            ITEMP, J, K, LASTRK, LSTICC, PVT, RK
                    102:       DOUBLE PRECISION   TEMP, TEMP2, TOL3Z
                    103:       COMPLEX*16         AKK
                    104: *     ..
                    105: *     .. External Subroutines ..
1.5       bertrand  106:       EXTERNAL           ZGEMM, ZGEMV, ZLARFG, ZSWAP
1.1       bertrand  107: *     ..
                    108: *     .. Intrinsic Functions ..
                    109:       INTRINSIC          ABS, DBLE, DCONJG, MAX, MIN, NINT, SQRT
                    110: *     ..
                    111: *     .. External Functions ..
                    112:       INTEGER            IDAMAX
                    113:       DOUBLE PRECISION   DLAMCH, DZNRM2
                    114:       EXTERNAL           IDAMAX, DLAMCH, DZNRM2
                    115: *     ..
                    116: *     .. Executable Statements ..
                    117: *
                    118:       LASTRK = MIN( M, N+OFFSET )
                    119:       LSTICC = 0
                    120:       K = 0
                    121:       TOL3Z = SQRT(DLAMCH('Epsilon'))
                    122: *
                    123: *     Beginning of while loop.
                    124: *
                    125:    10 CONTINUE
                    126:       IF( ( K.LT.NB ) .AND. ( LSTICC.EQ.0 ) ) THEN
                    127:          K = K + 1
                    128:          RK = OFFSET + K
                    129: *
                    130: *        Determine ith pivot column and swap if necessary
                    131: *
                    132:          PVT = ( K-1 ) + IDAMAX( N-K+1, VN1( K ), 1 )
                    133:          IF( PVT.NE.K ) THEN
                    134:             CALL ZSWAP( M, A( 1, PVT ), 1, A( 1, K ), 1 )
                    135:             CALL ZSWAP( K-1, F( PVT, 1 ), LDF, F( K, 1 ), LDF )
                    136:             ITEMP = JPVT( PVT )
                    137:             JPVT( PVT ) = JPVT( K )
                    138:             JPVT( K ) = ITEMP
                    139:             VN1( PVT ) = VN1( K )
                    140:             VN2( PVT ) = VN2( K )
                    141:          END IF
                    142: *
                    143: *        Apply previous Householder reflectors to column K:
                    144: *        A(RK:M,K) := A(RK:M,K) - A(RK:M,1:K-1)*F(K,1:K-1)'.
                    145: *
                    146:          IF( K.GT.1 ) THEN
                    147:             DO 20 J = 1, K - 1
                    148:                F( K, J ) = DCONJG( F( K, J ) )
                    149:    20       CONTINUE
                    150:             CALL ZGEMV( 'No transpose', M-RK+1, K-1, -CONE, A( RK, 1 ),
                    151:      $                  LDA, F( K, 1 ), LDF, CONE, A( RK, K ), 1 )
                    152:             DO 30 J = 1, K - 1
                    153:                F( K, J ) = DCONJG( F( K, J ) )
                    154:    30       CONTINUE
                    155:          END IF
                    156: *
                    157: *        Generate elementary reflector H(k).
                    158: *
                    159:          IF( RK.LT.M ) THEN
1.5       bertrand  160:             CALL ZLARFG( M-RK+1, A( RK, K ), A( RK+1, K ), 1, TAU( K ) )
1.1       bertrand  161:          ELSE
1.5       bertrand  162:             CALL ZLARFG( 1, A( RK, K ), A( RK, K ), 1, TAU( K ) )
1.1       bertrand  163:          END IF
                    164: *
                    165:          AKK = A( RK, K )
                    166:          A( RK, K ) = CONE
                    167: *
                    168: *        Compute Kth column of F:
                    169: *
                    170: *        Compute  F(K+1:N,K) := tau(K)*A(RK:M,K+1:N)'*A(RK:M,K).
                    171: *
                    172:          IF( K.LT.N ) THEN
                    173:             CALL ZGEMV( 'Conjugate transpose', M-RK+1, N-K, TAU( K ),
                    174:      $                  A( RK, K+1 ), LDA, A( RK, K ), 1, CZERO,
                    175:      $                  F( K+1, K ), 1 )
                    176:          END IF
                    177: *
                    178: *        Padding F(1:K,K) with zeros.
                    179: *
                    180:          DO 40 J = 1, K
                    181:             F( J, K ) = CZERO
                    182:    40    CONTINUE
                    183: *
                    184: *        Incremental updating of F:
                    185: *        F(1:N,K) := F(1:N,K) - tau(K)*F(1:N,1:K-1)*A(RK:M,1:K-1)'
                    186: *                    *A(RK:M,K).
                    187: *
                    188:          IF( K.GT.1 ) THEN
                    189:             CALL ZGEMV( 'Conjugate transpose', M-RK+1, K-1, -TAU( K ),
                    190:      $                  A( RK, 1 ), LDA, A( RK, K ), 1, CZERO,
                    191:      $                  AUXV( 1 ), 1 )
                    192: *
                    193:             CALL ZGEMV( 'No transpose', N, K-1, CONE, F( 1, 1 ), LDF,
                    194:      $                  AUXV( 1 ), 1, CONE, F( 1, K ), 1 )
                    195:          END IF
                    196: *
                    197: *        Update the current row of A:
                    198: *        A(RK,K+1:N) := A(RK,K+1:N) - A(RK,1:K)*F(K+1:N,1:K)'.
                    199: *
                    200:          IF( K.LT.N ) THEN
                    201:             CALL ZGEMM( 'No transpose', 'Conjugate transpose', 1, N-K,
                    202:      $                  K, -CONE, A( RK, 1 ), LDA, F( K+1, 1 ), LDF,
                    203:      $                  CONE, A( RK, K+1 ), LDA )
                    204:          END IF
                    205: *
                    206: *        Update partial column norms.
                    207: *
                    208:          IF( RK.LT.LASTRK ) THEN
                    209:             DO 50 J = K + 1, N
                    210:                IF( VN1( J ).NE.ZERO ) THEN
                    211: *
                    212: *                 NOTE: The following 4 lines follow from the analysis in
                    213: *                 Lapack Working Note 176.
                    214: *
                    215:                   TEMP = ABS( A( RK, J ) ) / VN1( J )
                    216:                   TEMP = MAX( ZERO, ( ONE+TEMP )*( ONE-TEMP ) )
                    217:                   TEMP2 = TEMP*( VN1( J ) / VN2( J ) )**2
                    218:                   IF( TEMP2 .LE. TOL3Z ) THEN
                    219:                      VN2( J ) = DBLE( LSTICC )
                    220:                      LSTICC = J
                    221:                   ELSE
                    222:                      VN1( J ) = VN1( J )*SQRT( TEMP )
                    223:                   END IF
                    224:                END IF
                    225:    50       CONTINUE
                    226:          END IF
                    227: *
                    228:          A( RK, K ) = AKK
                    229: *
                    230: *        End of while loop.
                    231: *
                    232:          GO TO 10
                    233:       END IF
                    234:       KB = K
                    235:       RK = OFFSET + KB
                    236: *
                    237: *     Apply the block reflector to the rest of the matrix:
                    238: *     A(OFFSET+KB+1:M,KB+1:N) := A(OFFSET+KB+1:M,KB+1:N) -
                    239: *                         A(OFFSET+KB+1:M,1:KB)*F(KB+1:N,1:KB)'.
                    240: *
                    241:       IF( KB.LT.MIN( N, M-OFFSET ) ) THEN
                    242:          CALL ZGEMM( 'No transpose', 'Conjugate transpose', M-RK, N-KB,
                    243:      $               KB, -CONE, A( RK+1, 1 ), LDA, F( KB+1, 1 ), LDF,
                    244:      $               CONE, A( RK+1, KB+1 ), LDA )
                    245:       END IF
                    246: *
                    247: *     Recomputation of difficult columns.
                    248: *
                    249:    60 CONTINUE
                    250:       IF( LSTICC.GT.0 ) THEN
                    251:          ITEMP = NINT( VN2( LSTICC ) )
                    252:          VN1( LSTICC ) = DZNRM2( M-RK, A( RK+1, LSTICC ), 1 )
                    253: *
                    254: *        NOTE: The computation of VN1( LSTICC ) relies on the fact that 
                    255: *        SNRM2 does not fail on vectors with norm below the value of
                    256: *        SQRT(DLAMCH('S')) 
                    257: *
                    258:          VN2( LSTICC ) = VN1( LSTICC )
                    259:          LSTICC = ITEMP
                    260:          GO TO 60
                    261:       END IF
                    262: *
                    263:       RETURN
                    264: *
                    265: *     End of ZLAQPS
                    266: *
                    267:       END

CVSweb interface <joel.bertrand@systella.fr>