LAPACK  3.5.0
LAPACK: Linear Algebra PACKage
sggsvd.f File Reference

Go to the source code of this file.

Functions/Subroutines

subroutine sggsvd (JOBU, JOBV, JOBQ, M, N, P, K, L, A, LDA, B, LDB, ALPHA, BETA, U, LDU, V, LDV, Q, LDQ, WORK, IWORK, INFO)
  SGGSVD computes the singular value decomposition (SVD) for OTHER matrices More...
 

Function/Subroutine Documentation

subroutine sggsvd ( character  JOBU,
character  JOBV,
character  JOBQ,
integer  M,
integer  N,
integer  P,
integer  K,
integer  L,
real, dimension( lda, * )  A,
integer  LDA,
real, dimension( ldb, * )  B,
integer  LDB,
real, dimension( * )  ALPHA,
real, dimension( * )  BETA,
real, dimension( ldu, * )  U,
integer  LDU,
real, dimension( ldv, * )  V,
integer  LDV,
real, dimension( ldq, * )  Q,
integer  LDQ,
real, dimension( * )  WORK,
integer, dimension( * )  IWORK,
integer  INFO 
)

SGGSVD computes the singular value decomposition (SVD) for OTHER matrices

Download SGGSVD + dependencies [TGZ] [ZIP] [TXT]

Purpose:
 SGGSVD computes the generalized singular value decomposition (GSVD)
 of an M-by-N real matrix A and P-by-N real matrix B:

       U**T*A*Q = D1*( 0 R ),    V**T*B*Q = D2*( 0 R )

 where U, V and Q are orthogonal matrices.
 Let K+L = the effective numerical rank of the matrix (A**T,B**T)**T,
 then R is a K+L-by-K+L nonsingular upper triangular matrix, D1 and
 D2 are M-by-(K+L) and P-by-(K+L) "diagonal" matrices and of the
 following structures, respectively:

 If M-K-L >= 0,

                     K  L
        D1 =     K ( I  0 )
                 L ( 0  C )
             M-K-L ( 0  0 )

                   K  L
        D2 =   L ( 0  S )
             P-L ( 0  0 )

                 N-K-L  K    L
   ( 0 R ) = K (  0   R11  R12 )
             L (  0    0   R22 )

 where

   C = diag( ALPHA(K+1), ... , ALPHA(K+L) ),
   S = diag( BETA(K+1),  ... , BETA(K+L) ),
   C**2 + S**2 = I.

   R is stored in A(1:K+L,N-K-L+1:N) on exit.

 If M-K-L < 0,

                   K M-K K+L-M
        D1 =   K ( I  0    0   )
             M-K ( 0  C    0   )

                     K M-K K+L-M
        D2 =   M-K ( 0  S    0  )
             K+L-M ( 0  0    I  )
               P-L ( 0  0    0  )

                    N-K-L  K   M-K  K+L-M
   ( 0 R ) =     K ( 0    R11  R12  R13  )
               M-K ( 0     0   R22  R23  )
             K+L-M ( 0     0    0   R33  )

 where

   C = diag( ALPHA(K+1), ... , ALPHA(M) ),
   S = diag( BETA(K+1),  ... , BETA(M) ),
   C**2 + S**2 = I.

   (R11 R12 R13 ) is stored in A(1:M, N-K-L+1:N), and R33 is stored
   ( 0  R22 R23 )
   in B(M-K+1:L,N+M-K-L+1:N) on exit.

 The routine computes C, S, R, and optionally the orthogonal
 transformation matrices U, V and Q.

 In particular, if B is an N-by-N nonsingular matrix, then the GSVD of
 A and B implicitly gives the SVD of A*inv(B):
                      A*inv(B) = U*(D1*inv(D2))*V**T.
 If ( A**T,B**T)**T  has orthonormal columns, then the GSVD of A and B is
 also equal to the CS decomposition of A and B. Furthermore, the GSVD
 can be used to derive the solution of the eigenvalue problem:
                      A**T*A x = lambda* B**T*B x.
 In some literature, the GSVD of A and B is presented in the form
                  U**T*A*X = ( 0 D1 ),   V**T*B*X = ( 0 D2 )
 where U and V are orthogonal and X is nonsingular, D1 and D2 are
 ``diagonal''.  The former GSVD form can be converted to the latter
 form by taking the nonsingular matrix X as

                      X = Q*( I   0    )
                            ( 0 inv(R) ).
Parameters
[in]JOBU
          JOBU is CHARACTER*1
          = 'U':  Orthogonal matrix U is computed;
          = 'N':  U is not computed.
[in]JOBV
          JOBV is CHARACTER*1
          = 'V':  Orthogonal matrix V is computed;
          = 'N':  V is not computed.
[in]JOBQ
          JOBQ is CHARACTER*1
          = 'Q':  Orthogonal matrix Q is computed;
          = 'N':  Q is not computed.
[in]M
          M is INTEGER
          The number of rows of the matrix A.  M >= 0.
[in]N
          N is INTEGER
          The number of columns of the matrices A and B.  N >= 0.
[in]P
          P is INTEGER
          The number of rows of the matrix B.  P >= 0.
[out]K
          K is INTEGER
[out]L
          L is INTEGER

          On exit, K and L specify the dimension of the subblocks
          described in Purpose.
          K + L = effective numerical rank of (A**T,B**T)**T.
[in,out]A
          A is REAL array, dimension (LDA,N)
          On entry, the M-by-N matrix A.
          On exit, A contains the triangular matrix R, or part of R.
          See Purpose for details.
[in]LDA
          LDA is INTEGER
          The leading dimension of the array A. LDA >= max(1,M).
[in,out]B
          B is REAL array, dimension (LDB,N)
          On entry, the P-by-N matrix B.
          On exit, B contains the triangular matrix R if M-K-L < 0.
          See Purpose for details.
[in]LDB
          LDB is INTEGER
          The leading dimension of the array B. LDB >= max(1,P).
[out]ALPHA
          ALPHA is REAL array, dimension (N)
[out]BETA
          BETA is REAL array, dimension (N)

          On exit, ALPHA and BETA contain the generalized singular
          value pairs of A and B;
            ALPHA(1:K) = 1,
            BETA(1:K)  = 0,
          and if M-K-L >= 0,
            ALPHA(K+1:K+L) = C,
            BETA(K+1:K+L)  = S,
          or if M-K-L < 0,
            ALPHA(K+1:M)=C, ALPHA(M+1:K+L)=0
            BETA(K+1:M) =S, BETA(M+1:K+L) =1
          and
            ALPHA(K+L+1:N) = 0
            BETA(K+L+1:N)  = 0
[out]U
          U is REAL array, dimension (LDU,M)
          If JOBU = 'U', U contains the M-by-M orthogonal matrix U.
          If JOBU = 'N', U is not referenced.
[in]LDU
          LDU is INTEGER
          The leading dimension of the array U. LDU >= max(1,M) if
          JOBU = 'U'; LDU >= 1 otherwise.
[out]V
          V is REAL array, dimension (LDV,P)
          If JOBV = 'V', V contains the P-by-P orthogonal matrix V.
          If JOBV = 'N', V is not referenced.
[in]LDV
          LDV is INTEGER
          The leading dimension of the array V. LDV >= max(1,P) if
          JOBV = 'V'; LDV >= 1 otherwise.
[out]Q
          Q is REAL array, dimension (LDQ,N)
          If JOBQ = 'Q', Q contains the N-by-N orthogonal matrix Q.
          If JOBQ = 'N', Q is not referenced.
[in]LDQ
          LDQ is INTEGER
          The leading dimension of the array Q. LDQ >= max(1,N) if
          JOBQ = 'Q'; LDQ >= 1 otherwise.
[out]WORK
          WORK is REAL array,
                      dimension (max(3*N,M,P)+N)
[out]IWORK
          IWORK is INTEGER array, dimension (N)
          On exit, IWORK stores the sorting information. More
          precisely, the following loop will sort ALPHA
             for I = K+1, min(M,K+L)
                 swap ALPHA(I) and ALPHA(IWORK(I))
             endfor
          such that ALPHA(1) >= ALPHA(2) >= ... >= ALPHA(N).
[out]INFO
          INFO is INTEGER
          = 0:  successful exit
          < 0:  if INFO = -i, the i-th argument had an illegal value.
          > 0:  if INFO = 1, the Jacobi-type procedure failed to
                converge.  For further details, see subroutine STGSJA.
Internal Parameters:
  TOLA    REAL
  TOLB    REAL
          TOLA and TOLB are the thresholds to determine the effective
          rank of (A**T,B**T)**T. Generally, they are set to
                   TOLA = MAX(M,N)*norm(A)*MACHEPS,
                   TOLB = MAX(P,N)*norm(B)*MACHEPS.
          The size of TOLA and TOLB may affect the size of backward
          errors of the decomposition.
Author
Univ. of Tennessee
Univ. of California Berkeley
Univ. of Colorado Denver
NAG Ltd.
Date
November 2011
Contributors:
Ming Gu and Huan Ren, Computer Science Division, University of California at Berkeley, USA

Definition at line 334 of file sggsvd.f.

334 *
335 * -- LAPACK driver routine (version 3.4.0) --
336 * -- LAPACK is a software package provided by Univ. of Tennessee, --
337 * -- Univ. of California Berkeley, Univ. of Colorado Denver and NAG Ltd..--
338 * November 2011
339 *
340 * .. Scalar Arguments ..
341  CHARACTER jobq, jobu, jobv
342  INTEGER info, k, l, lda, ldb, ldq, ldu, ldv, m, n, p
343 * ..
344 * .. Array Arguments ..
345  INTEGER iwork( * )
346  REAL a( lda, * ), alpha( * ), b( ldb, * ),
347  $ beta( * ), q( ldq, * ), u( ldu, * ),
348  $ v( ldv, * ), work( * )
349 * ..
350 *
351 * =====================================================================
352 *
353 * .. Local Scalars ..
354  LOGICAL wantq, wantu, wantv
355  INTEGER i, ibnd, isub, j, ncycle
356  REAL anorm, bnorm, smax, temp, tola, tolb, ulp, unfl
357 * ..
358 * .. External Functions ..
359  LOGICAL lsame
360  REAL slamch, slange
361  EXTERNAL lsame, slamch, slange
362 * ..
363 * .. External Subroutines ..
364  EXTERNAL scopy, sggsvp, stgsja, xerbla
365 * ..
366 * .. Intrinsic Functions ..
367  INTRINSIC max, min
368 * ..
369 * .. Executable Statements ..
370 *
371 * Test the input parameters
372 *
373  wantu = lsame( jobu, 'U' )
374  wantv = lsame( jobv, 'V' )
375  wantq = lsame( jobq, 'Q' )
376 *
377  info = 0
378  IF( .NOT.( wantu .OR. lsame( jobu, 'N' ) ) ) THEN
379  info = -1
380  ELSE IF( .NOT.( wantv .OR. lsame( jobv, 'N' ) ) ) THEN
381  info = -2
382  ELSE IF( .NOT.( wantq .OR. lsame( jobq, 'N' ) ) ) THEN
383  info = -3
384  ELSE IF( m.LT.0 ) THEN
385  info = -4
386  ELSE IF( n.LT.0 ) THEN
387  info = -5
388  ELSE IF( p.LT.0 ) THEN
389  info = -6
390  ELSE IF( lda.LT.max( 1, m ) ) THEN
391  info = -10
392  ELSE IF( ldb.LT.max( 1, p ) ) THEN
393  info = -12
394  ELSE IF( ldu.LT.1 .OR. ( wantu .AND. ldu.LT.m ) ) THEN
395  info = -16
396  ELSE IF( ldv.LT.1 .OR. ( wantv .AND. ldv.LT.p ) ) THEN
397  info = -18
398  ELSE IF( ldq.LT.1 .OR. ( wantq .AND. ldq.LT.n ) ) THEN
399  info = -20
400  END IF
401  IF( info.NE.0 ) THEN
402  CALL xerbla( 'SGGSVD', -info )
403  RETURN
404  END IF
405 *
406 * Compute the Frobenius norm of matrices A and B
407 *
408  anorm = slange( '1', m, n, a, lda, work )
409  bnorm = slange( '1', p, n, b, ldb, work )
410 *
411 * Get machine precision and set up threshold for determining
412 * the effective numerical rank of the matrices A and B.
413 *
414  ulp = slamch( 'Precision' )
415  unfl = slamch( 'Safe Minimum' )
416  tola = max( m, n )*max( anorm, unfl )*ulp
417  tolb = max( p, n )*max( bnorm, unfl )*ulp
418 *
419 * Preprocessing
420 *
421  CALL sggsvp( jobu, jobv, jobq, m, p, n, a, lda, b, ldb, tola,
422  $ tolb, k, l, u, ldu, v, ldv, q, ldq, iwork, work,
423  $ work( n+1 ), info )
424 *
425 * Compute the GSVD of two upper "triangular" matrices
426 *
427  CALL stgsja( jobu, jobv, jobq, m, p, n, k, l, a, lda, b, ldb,
428  $ tola, tolb, alpha, beta, u, ldu, v, ldv, q, ldq,
429  $ work, ncycle, info )
430 *
431 * Sort the singular values and store the pivot indices in IWORK
432 * Copy ALPHA to WORK, then sort ALPHA in WORK
433 *
434  CALL scopy( n, alpha, 1, work, 1 )
435  ibnd = min( l, m-k )
436  DO 20 i = 1, ibnd
437 *
438 * Scan for largest ALPHA(K+I)
439 *
440  isub = i
441  smax = work( k+i )
442  DO 10 j = i + 1, ibnd
443  temp = work( k+j )
444  IF( temp.GT.smax ) THEN
445  isub = j
446  smax = temp
447  END IF
448  10 CONTINUE
449  IF( isub.NE.i ) THEN
450  work( k+isub ) = work( k+i )
451  work( k+i ) = smax
452  iwork( k+i ) = k + isub
453  ELSE
454  iwork( k+i ) = k + i
455  END IF
456  20 CONTINUE
457 *
458  RETURN
459 *
460 * End of SGGSVD
461 *
real function slange(NORM, M, N, A, LDA, WORK)
SLANGE returns the value of the 1-norm, Frobenius norm, infinity-norm, or the largest absolute value ...
Definition: slange.f:116
subroutine xerbla(SRNAME, INFO)
XERBLA
Definition: xerbla.f:62
subroutine stgsja(JOBU, JOBV, JOBQ, M, P, N, K, L, A, LDA, B, LDB, TOLA, TOLB, ALPHA, BETA, U, LDU, V, LDV, Q, LDQ, WORK, NCYCLE, INFO)
STGSJA
Definition: stgsja.f:380
subroutine scopy(N, SX, INCX, SY, INCY)
SCOPY
Definition: scopy.f:53
logical function lsame(CA, CB)
LSAME
Definition: lsame.f:55
real function slamch(CMACH)
SLAMCH
Definition: slamch.f:69
subroutine sggsvp(JOBU, JOBV, JOBQ, M, P, N, A, LDA, B, LDB, TOLA, TOLB, K, L, U, LDU, V, LDV, Q, LDQ, IWORK, TAU, WORK, INFO)
SGGSVP
Definition: sggsvp.f:256

Here is the call graph for this function:

Here is the caller graph for this function: