Annotation of rpl/lapack/lapack/zggsvd3.f, revision 1.3

1.1       bertrand    1: *> \brief <b> ZGGSVD3 computes the singular value decomposition (SVD) for OTHER matrices</b>
                      2: *
                      3: *  =========== DOCUMENTATION ===========
                      4: *
                      5: * Online html documentation available at 
                      6: *            http://www.netlib.org/lapack/explore-html/ 
                      7: *
                      8: *> \htmlonly
                      9: *> Download ZGGSVD3 + dependencies 
                     10: *> <a href="http://www.netlib.org/cgi-bin/netlibfiles.tgz?format=tgz&filename=/lapack/lapack_routine/zggsvd3.f"> 
                     11: *> [TGZ]</a> 
                     12: *> <a href="http://www.netlib.org/cgi-bin/netlibfiles.zip?format=zip&filename=/lapack/lapack_routine/zggsvd3.f"> 
                     13: *> [ZIP]</a> 
                     14: *> <a href="http://www.netlib.org/cgi-bin/netlibfiles.txt?format=txt&filename=/lapack/lapack_routine/zggsvd3.f"> 
                     15: *> [TXT]</a>
                     16: *> \endhtmlonly 
                     17: *
                     18: *  Definition:
                     19: *  ===========
                     20: *
                     21: *       SUBROUTINE ZGGSVD3( JOBU, JOBV, JOBQ, M, N, P, K, L, A, LDA, B,
                     22: *                           LDB, ALPHA, BETA, U, LDU, V, LDV, Q, LDQ, WORK,
                     23: *                           LWORK, RWORK, IWORK, INFO )
                     24: * 
                     25: *       .. Scalar Arguments ..
                     26: *       CHARACTER          JOBQ, JOBU, JOBV
                     27: *       INTEGER            INFO, K, L, LDA, LDB, LDQ, LDU, LDV, M, N, P, LWORK
                     28: *       ..
                     29: *       .. Array Arguments ..
                     30: *       INTEGER            IWORK( * )
                     31: *       DOUBLE PRECISION   ALPHA( * ), BETA( * ), RWORK( * )
                     32: *       COMPLEX*16         A( LDA, * ), B( LDB, * ), Q( LDQ, * ),
                     33: *      $                   U( LDU, * ), V( LDV, * ), WORK( * )
                     34: *       ..
                     35: *  
                     36: *
                     37: *> \par Purpose:
                     38: *  =============
                     39: *>
                     40: *> \verbatim
                     41: *>
                     42: *> ZGGSVD3 computes the generalized singular value decomposition (GSVD)
                     43: *> of an M-by-N complex matrix A and P-by-N complex matrix B:
                     44: *>
                     45: *>       U**H*A*Q = D1*( 0 R ),    V**H*B*Q = D2*( 0 R )
                     46: *>
                     47: *> where U, V and Q are unitary matrices.
                     48: *> Let K+L = the effective numerical rank of the
                     49: *> matrix (A**H,B**H)**H, then R is a (K+L)-by-(K+L) nonsingular upper
                     50: *> triangular matrix, D1 and D2 are M-by-(K+L) and P-by-(K+L) "diagonal"
                     51: *> matrices and of the following structures, respectively:
                     52: *>
                     53: *> If M-K-L >= 0,
                     54: *>
                     55: *>                     K  L
                     56: *>        D1 =     K ( I  0 )
                     57: *>                 L ( 0  C )
                     58: *>             M-K-L ( 0  0 )
                     59: *>
                     60: *>                   K  L
                     61: *>        D2 =   L ( 0  S )
                     62: *>             P-L ( 0  0 )
                     63: *>
                     64: *>                 N-K-L  K    L
                     65: *>   ( 0 R ) = K (  0   R11  R12 )
                     66: *>             L (  0    0   R22 )
                     67: *> where
                     68: *>
                     69: *>   C = diag( ALPHA(K+1), ... , ALPHA(K+L) ),
                     70: *>   S = diag( BETA(K+1),  ... , BETA(K+L) ),
                     71: *>   C**2 + S**2 = I.
                     72: *>
                     73: *>   R is stored in A(1:K+L,N-K-L+1:N) on exit.
                     74: *>
                     75: *> If M-K-L < 0,
                     76: *>
                     77: *>                   K M-K K+L-M
                     78: *>        D1 =   K ( I  0    0   )
                     79: *>             M-K ( 0  C    0   )
                     80: *>
                     81: *>                     K M-K K+L-M
                     82: *>        D2 =   M-K ( 0  S    0  )
                     83: *>             K+L-M ( 0  0    I  )
                     84: *>               P-L ( 0  0    0  )
                     85: *>
                     86: *>                    N-K-L  K   M-K  K+L-M
                     87: *>   ( 0 R ) =     K ( 0    R11  R12  R13  )
                     88: *>               M-K ( 0     0   R22  R23  )
                     89: *>             K+L-M ( 0     0    0   R33  )
                     90: *>
                     91: *> where
                     92: *>
                     93: *>   C = diag( ALPHA(K+1), ... , ALPHA(M) ),
                     94: *>   S = diag( BETA(K+1),  ... , BETA(M) ),
                     95: *>   C**2 + S**2 = I.
                     96: *>
                     97: *>   (R11 R12 R13 ) is stored in A(1:M, N-K-L+1:N), and R33 is stored
                     98: *>   ( 0  R22 R23 )
                     99: *>   in B(M-K+1:L,N+M-K-L+1:N) on exit.
                    100: *>
                    101: *> The routine computes C, S, R, and optionally the unitary
                    102: *> transformation matrices U, V and Q.
                    103: *>
                    104: *> In particular, if B is an N-by-N nonsingular matrix, then the GSVD of
                    105: *> A and B implicitly gives the SVD of A*inv(B):
                    106: *>                      A*inv(B) = U*(D1*inv(D2))*V**H.
                    107: *> If ( A**H,B**H)**H has orthonormal columns, then the GSVD of A and B is also
                    108: *> equal to the CS decomposition of A and B. Furthermore, the GSVD can
                    109: *> be used to derive the solution of the eigenvalue problem:
                    110: *>                      A**H*A x = lambda* B**H*B x.
                    111: *> In some literature, the GSVD of A and B is presented in the form
                    112: *>                  U**H*A*X = ( 0 D1 ),   V**H*B*X = ( 0 D2 )
                    113: *> where U and V are orthogonal and X is nonsingular, and D1 and D2 are
                    114: *> ``diagonal''.  The former GSVD form can be converted to the latter
                    115: *> form by taking the nonsingular matrix X as
                    116: *>
                    117: *>                       X = Q*(  I   0    )
                    118: *>                             (  0 inv(R) )
                    119: *> \endverbatim
                    120: *
                    121: *  Arguments:
                    122: *  ==========
                    123: *
                    124: *> \param[in] JOBU
                    125: *> \verbatim
                    126: *>          JOBU is CHARACTER*1
                    127: *>          = 'U':  Unitary matrix U is computed;
                    128: *>          = 'N':  U is not computed.
                    129: *> \endverbatim
                    130: *>
                    131: *> \param[in] JOBV
                    132: *> \verbatim
                    133: *>          JOBV is CHARACTER*1
                    134: *>          = 'V':  Unitary matrix V is computed;
                    135: *>          = 'N':  V is not computed.
                    136: *> \endverbatim
                    137: *>
                    138: *> \param[in] JOBQ
                    139: *> \verbatim
                    140: *>          JOBQ is CHARACTER*1
                    141: *>          = 'Q':  Unitary matrix Q is computed;
                    142: *>          = 'N':  Q is not computed.
                    143: *> \endverbatim
                    144: *>
                    145: *> \param[in] M
                    146: *> \verbatim
                    147: *>          M is INTEGER
                    148: *>          The number of rows of the matrix A.  M >= 0.
                    149: *> \endverbatim
                    150: *>
                    151: *> \param[in] N
                    152: *> \verbatim
                    153: *>          N is INTEGER
                    154: *>          The number of columns of the matrices A and B.  N >= 0.
                    155: *> \endverbatim
                    156: *>
                    157: *> \param[in] P
                    158: *> \verbatim
                    159: *>          P is INTEGER
                    160: *>          The number of rows of the matrix B.  P >= 0.
                    161: *> \endverbatim
                    162: *>
                    163: *> \param[out] K
                    164: *> \verbatim
                    165: *>          K is INTEGER
                    166: *> \endverbatim
                    167: *>
                    168: *> \param[out] L
                    169: *> \verbatim
                    170: *>          L is INTEGER
                    171: *>
                    172: *>          On exit, K and L specify the dimension of the subblocks
                    173: *>          described in Purpose.
                    174: *>          K + L = effective numerical rank of (A**H,B**H)**H.
                    175: *> \endverbatim
                    176: *>
                    177: *> \param[in,out] A
                    178: *> \verbatim
                    179: *>          A is COMPLEX*16 array, dimension (LDA,N)
                    180: *>          On entry, the M-by-N matrix A.
                    181: *>          On exit, A contains the triangular matrix R, or part of R.
                    182: *>          See Purpose for details.
                    183: *> \endverbatim
                    184: *>
                    185: *> \param[in] LDA
                    186: *> \verbatim
                    187: *>          LDA is INTEGER
                    188: *>          The leading dimension of the array A. LDA >= max(1,M).
                    189: *> \endverbatim
                    190: *>
                    191: *> \param[in,out] B
                    192: *> \verbatim
                    193: *>          B is COMPLEX*16 array, dimension (LDB,N)
                    194: *>          On entry, the P-by-N matrix B.
                    195: *>          On exit, B contains part of the triangular matrix R if
                    196: *>          M-K-L < 0.  See Purpose for details.
                    197: *> \endverbatim
                    198: *>
                    199: *> \param[in] LDB
                    200: *> \verbatim
                    201: *>          LDB is INTEGER
                    202: *>          The leading dimension of the array B. LDB >= max(1,P).
                    203: *> \endverbatim
                    204: *>
                    205: *> \param[out] ALPHA
                    206: *> \verbatim
                    207: *>          ALPHA is DOUBLE PRECISION array, dimension (N)
                    208: *> \endverbatim
                    209: *>
                    210: *> \param[out] BETA
                    211: *> \verbatim
                    212: *>          BETA is DOUBLE PRECISION array, dimension (N)
                    213: *>
                    214: *>          On exit, ALPHA and BETA contain the generalized singular
                    215: *>          value pairs of A and B;
                    216: *>            ALPHA(1:K) = 1,
                    217: *>            BETA(1:K)  = 0,
                    218: *>          and if M-K-L >= 0,
                    219: *>            ALPHA(K+1:K+L) = C,
                    220: *>            BETA(K+1:K+L)  = S,
                    221: *>          or if M-K-L < 0,
                    222: *>            ALPHA(K+1:M)=C, ALPHA(M+1:K+L)=0
                    223: *>            BETA(K+1:M) =S, BETA(M+1:K+L) =1
                    224: *>          and
                    225: *>            ALPHA(K+L+1:N) = 0
                    226: *>            BETA(K+L+1:N)  = 0
                    227: *> \endverbatim
                    228: *>
                    229: *> \param[out] U
                    230: *> \verbatim
                    231: *>          U is COMPLEX*16 array, dimension (LDU,M)
                    232: *>          If JOBU = 'U', U contains the M-by-M unitary matrix U.
                    233: *>          If JOBU = 'N', U is not referenced.
                    234: *> \endverbatim
                    235: *>
                    236: *> \param[in] LDU
                    237: *> \verbatim
                    238: *>          LDU is INTEGER
                    239: *>          The leading dimension of the array U. LDU >= max(1,M) if
                    240: *>          JOBU = 'U'; LDU >= 1 otherwise.
                    241: *> \endverbatim
                    242: *>
                    243: *> \param[out] V
                    244: *> \verbatim
                    245: *>          V is COMPLEX*16 array, dimension (LDV,P)
                    246: *>          If JOBV = 'V', V contains the P-by-P unitary matrix V.
                    247: *>          If JOBV = 'N', V is not referenced.
                    248: *> \endverbatim
                    249: *>
                    250: *> \param[in] LDV
                    251: *> \verbatim
                    252: *>          LDV is INTEGER
                    253: *>          The leading dimension of the array V. LDV >= max(1,P) if
                    254: *>          JOBV = 'V'; LDV >= 1 otherwise.
                    255: *> \endverbatim
                    256: *>
                    257: *> \param[out] Q
                    258: *> \verbatim
                    259: *>          Q is COMPLEX*16 array, dimension (LDQ,N)
                    260: *>          If JOBQ = 'Q', Q contains the N-by-N unitary matrix Q.
                    261: *>          If JOBQ = 'N', Q is not referenced.
                    262: *> \endverbatim
                    263: *>
                    264: *> \param[in] LDQ
                    265: *> \verbatim
                    266: *>          LDQ is INTEGER
                    267: *>          The leading dimension of the array Q. LDQ >= max(1,N) if
                    268: *>          JOBQ = 'Q'; LDQ >= 1 otherwise.
                    269: *> \endverbatim
                    270: *>
                    271: *> \param[out] WORK
                    272: *> \verbatim
                    273: *>          WORK is COMPLEX*16 array, dimension (MAX(1,LWORK))
                    274: *>          On exit, if INFO = 0, WORK(1) returns the optimal LWORK.
                    275: *> \endverbatim
                    276: *>
                    277: *> \param[in] LWORK
                    278: *> \verbatim
                    279: *>          LWORK is INTEGER
                    280: *>          The dimension of the array WORK.
                    281: *>
                    282: *>          If LWORK = -1, then a workspace query is assumed; the routine
                    283: *>          only calculates the optimal size of the WORK array, returns
                    284: *>          this value as the first entry of the WORK array, and no error
                    285: *>          message related to LWORK is issued by XERBLA.
                    286: *> \endverbatim
                    287: *>
                    288: *> \param[out] RWORK
                    289: *> \verbatim
                    290: *>          RWORK is DOUBLE PRECISION array, dimension (2*N)
                    291: *> \endverbatim
                    292: *>
                    293: *> \param[out] IWORK
                    294: *> \verbatim
                    295: *>          IWORK is INTEGER array, dimension (N)
                    296: *>          On exit, IWORK stores the sorting information. More
                    297: *>          precisely, the following loop will sort ALPHA
                    298: *>             for I = K+1, min(M,K+L)
                    299: *>                 swap ALPHA(I) and ALPHA(IWORK(I))
                    300: *>             endfor
                    301: *>          such that ALPHA(1) >= ALPHA(2) >= ... >= ALPHA(N).
                    302: *> \endverbatim
                    303: *>
                    304: *> \param[out] INFO
                    305: *> \verbatim
                    306: *>          INFO is INTEGER
                    307: *>          = 0:  successful exit.
                    308: *>          < 0:  if INFO = -i, the i-th argument had an illegal value.
                    309: *>          > 0:  if INFO = 1, the Jacobi-type procedure failed to
                    310: *>                converge.  For further details, see subroutine ZTGSJA.
                    311: *> \endverbatim
                    312: *
                    313: *> \par Internal Parameters:
                    314: *  =========================
                    315: *>
                    316: *> \verbatim
                    317: *>  TOLA    DOUBLE PRECISION
                    318: *>  TOLB    DOUBLE PRECISION
                    319: *>          TOLA and TOLB are the thresholds to determine the effective
                    320: *>          rank of (A**H,B**H)**H. Generally, they are set to
                    321: *>                   TOLA = MAX(M,N)*norm(A)*MACHEPS,
                    322: *>                   TOLB = MAX(P,N)*norm(B)*MACHEPS.
                    323: *>          The size of TOLA and TOLB may affect the size of backward
                    324: *>          errors of the decomposition.
                    325: *> \endverbatim
                    326: *
                    327: *  Authors:
                    328: *  ========
                    329: *
                    330: *> \author Univ. of Tennessee 
                    331: *> \author Univ. of California Berkeley 
                    332: *> \author Univ. of Colorado Denver 
                    333: *> \author NAG Ltd. 
                    334: *
                    335: *> \date August 2015
                    336: *
                    337: *> \ingroup complex16OTHERsing
                    338: *
                    339: *> \par Contributors:
                    340: *  ==================
                    341: *>
                    342: *>     Ming Gu and Huan Ren, Computer Science Division, University of
                    343: *>     California at Berkeley, USA
                    344: *>
                    345: *
                    346: *> \par Further Details:
                    347: *  =====================
                    348: *>
                    349: *>  ZGGSVD3 replaces the deprecated subroutine ZGGSVD.
                    350: *>
                    351: *  =====================================================================
                    352:       SUBROUTINE ZGGSVD3( JOBU, JOBV, JOBQ, M, N, P, K, L, A, LDA, B,
                    353:      $                    LDB, ALPHA, BETA, U, LDU, V, LDV, Q, LDQ,
                    354:      $                    WORK, LWORK, RWORK, IWORK, INFO )
                    355: *
1.2       bertrand  356: *  -- LAPACK driver routine (version 3.6.1) --
1.1       bertrand  357: *  -- LAPACK is a software package provided by Univ. of Tennessee,    --
                    358: *  -- Univ. of California Berkeley, Univ. of Colorado Denver and NAG Ltd..--
                    359: *     August 2015
                    360: *
                    361: *     .. Scalar Arguments ..
                    362:       CHARACTER          JOBQ, JOBU, JOBV
                    363:       INTEGER            INFO, K, L, LDA, LDB, LDQ, LDU, LDV, M, N, P,
                    364:      $                   LWORK
                    365: *     ..
                    366: *     .. Array Arguments ..
                    367:       INTEGER            IWORK( * )
                    368:       DOUBLE PRECISION   ALPHA( * ), BETA( * ), RWORK( * )
                    369:       COMPLEX*16         A( LDA, * ), B( LDB, * ), Q( LDQ, * ),
                    370:      $                   U( LDU, * ), V( LDV, * ), WORK( * )
                    371: *     ..
                    372: *
                    373: *  =====================================================================
                    374: *
                    375: *     .. Local Scalars ..
                    376:       LOGICAL            WANTQ, WANTU, WANTV, LQUERY
                    377:       INTEGER            I, IBND, ISUB, J, NCYCLE, LWKOPT
                    378:       DOUBLE PRECISION   ANORM, BNORM, SMAX, TEMP, TOLA, TOLB, ULP, UNFL
                    379: *     ..
                    380: *     .. External Functions ..
                    381:       LOGICAL            LSAME
                    382:       DOUBLE PRECISION   DLAMCH, ZLANGE
                    383:       EXTERNAL           LSAME, DLAMCH, ZLANGE
                    384: *     ..
                    385: *     .. External Subroutines ..
1.2       bertrand  386:       EXTERNAL           DCOPY, XERBLA, ZGGSVP3, ZTGSJA
1.1       bertrand  387: *     ..
                    388: *     .. Intrinsic Functions ..
                    389:       INTRINSIC          MAX, MIN
                    390: *     ..
                    391: *     .. Executable Statements ..
                    392: *
                    393: *     Decode and test the input parameters
                    394: *
                    395:       WANTU = LSAME( JOBU, 'U' )
                    396:       WANTV = LSAME( JOBV, 'V' )
                    397:       WANTQ = LSAME( JOBQ, 'Q' )
                    398:       LQUERY = ( LWORK.EQ.-1 )
                    399:       LWKOPT = 1
                    400: *
                    401: *     Test the input arguments
                    402: *
                    403:       INFO = 0
                    404:       IF( .NOT.( WANTU .OR. LSAME( JOBU, 'N' ) ) ) THEN
                    405:          INFO = -1
                    406:       ELSE IF( .NOT.( WANTV .OR. LSAME( JOBV, 'N' ) ) ) THEN
                    407:          INFO = -2
                    408:       ELSE IF( .NOT.( WANTQ .OR. LSAME( JOBQ, 'N' ) ) ) THEN
                    409:          INFO = -3
                    410:       ELSE IF( M.LT.0 ) THEN
                    411:          INFO = -4
                    412:       ELSE IF( N.LT.0 ) THEN
                    413:          INFO = -5
                    414:       ELSE IF( P.LT.0 ) THEN
                    415:          INFO = -6
                    416:       ELSE IF( LDA.LT.MAX( 1, M ) ) THEN
                    417:          INFO = -10
                    418:       ELSE IF( LDB.LT.MAX( 1, P ) ) THEN
                    419:          INFO = -12
                    420:       ELSE IF( LDU.LT.1 .OR. ( WANTU .AND. LDU.LT.M ) ) THEN
                    421:          INFO = -16
                    422:       ELSE IF( LDV.LT.1 .OR. ( WANTV .AND. LDV.LT.P ) ) THEN
                    423:          INFO = -18
                    424:       ELSE IF( LDQ.LT.1 .OR. ( WANTQ .AND. LDQ.LT.N ) ) THEN
                    425:          INFO = -20
                    426:       ELSE IF( LWORK.LT.1 .AND. .NOT.LQUERY ) THEN
                    427:          INFO = -24
                    428:       END IF
                    429: *
                    430: *     Compute workspace
                    431: *
                    432:       IF( INFO.EQ.0 ) THEN
                    433:          CALL ZGGSVP3( JOBU, JOBV, JOBQ, M, P, N, A, LDA, B, LDB, TOLA,
                    434:      $                 TOLB, K, L, U, LDU, V, LDV, Q, LDQ, IWORK, RWORK,
                    435:      $                 WORK, WORK, -1, INFO )
                    436:          LWKOPT = N + INT( WORK( 1 ) )
                    437:          LWKOPT = MAX( 2*N, LWKOPT )
                    438:          LWKOPT = MAX( 1, LWKOPT )
                    439:          WORK( 1 ) = DCMPLX( LWKOPT )
                    440:       END IF
                    441: *
                    442:       IF( INFO.NE.0 ) THEN
                    443:          CALL XERBLA( 'ZGGSVD3', -INFO )
                    444:          RETURN
                    445:       END IF
                    446:       IF( LQUERY ) THEN
                    447:          RETURN
                    448:       ENDIF
                    449: *
                    450: *     Compute the Frobenius norm of matrices A and B
                    451: *
                    452:       ANORM = ZLANGE( '1', M, N, A, LDA, RWORK )
                    453:       BNORM = ZLANGE( '1', P, N, B, LDB, RWORK )
                    454: *
                    455: *     Get machine precision and set up threshold for determining
                    456: *     the effective numerical rank of the matrices A and B.
                    457: *
                    458:       ULP = DLAMCH( 'Precision' )
                    459:       UNFL = DLAMCH( 'Safe Minimum' )
                    460:       TOLA = MAX( M, N )*MAX( ANORM, UNFL )*ULP
                    461:       TOLB = MAX( P, N )*MAX( BNORM, UNFL )*ULP
                    462: *
                    463:       CALL ZGGSVP3( JOBU, JOBV, JOBQ, M, P, N, A, LDA, B, LDB, TOLA,
                    464:      $              TOLB, K, L, U, LDU, V, LDV, Q, LDQ, IWORK, RWORK,
                    465:      $              WORK, WORK( N+1 ), LWORK-N, INFO )
                    466: *
                    467: *     Compute the GSVD of two upper "triangular" matrices
                    468: *
                    469:       CALL ZTGSJA( JOBU, JOBV, JOBQ, M, P, N, K, L, A, LDA, B, LDB,
                    470:      $             TOLA, TOLB, ALPHA, BETA, U, LDU, V, LDV, Q, LDQ,
                    471:      $             WORK, NCYCLE, INFO )
                    472: *
                    473: *     Sort the singular values and store the pivot indices in IWORK
                    474: *     Copy ALPHA to RWORK, then sort ALPHA in RWORK
                    475: *
                    476:       CALL DCOPY( N, ALPHA, 1, RWORK, 1 )
                    477:       IBND = MIN( L, M-K )
                    478:       DO 20 I = 1, IBND
                    479: *
                    480: *        Scan for largest ALPHA(K+I)
                    481: *
                    482:          ISUB = I
                    483:          SMAX = RWORK( K+I )
                    484:          DO 10 J = I + 1, IBND
                    485:             TEMP = RWORK( K+J )
                    486:             IF( TEMP.GT.SMAX ) THEN
                    487:                ISUB = J
                    488:                SMAX = TEMP
                    489:             END IF
                    490:    10    CONTINUE
                    491:          IF( ISUB.NE.I ) THEN
                    492:             RWORK( K+ISUB ) = RWORK( K+I )
                    493:             RWORK( K+I ) = SMAX
                    494:             IWORK( K+I ) = K + ISUB
                    495:          ELSE
                    496:             IWORK( K+I ) = K + I
                    497:          END IF
                    498:    20 CONTINUE
                    499: *
                    500:       WORK( 1 ) = DCMPLX( LWKOPT )
                    501:       RETURN
                    502: *
                    503: *     End of ZGGSVD3
                    504: *
                    505:       END

CVSweb interface <joel.bertrand@systella.fr>