summaryrefslogtreecommitdiff
path: root/TESTING/LIN/dsyt01_aa.f
diff options
context:
space:
mode:
Diffstat (limited to 'TESTING/LIN/dsyt01_aa.f')
-rw-r--r--TESTING/LIN/dsyt01_aa.f263
1 files changed, 263 insertions, 0 deletions
diff --git a/TESTING/LIN/dsyt01_aa.f b/TESTING/LIN/dsyt01_aa.f
new file mode 100644
index 00000000..bc30df38
--- /dev/null
+++ b/TESTING/LIN/dsyt01_aa.f
@@ -0,0 +1,263 @@
+*> \brief \b DSYT01
+*
+* =========== DOCUMENTATION ===========
+*
+* Online html documentation available at
+* http://www.netlib.org/lapack/explore-html/
+*
+* Definition:
+* ===========
+*
+* SUBROUTINE DSYT01( UPLO, N, A, LDA, AFAC, LDAFAC, IPIV, C, LDC,
+* RWORK, RESID )
+*
+* .. Scalar Arguments ..
+* CHARACTER UPLO
+* INTEGER LDA, LDAFAC, LDC, N
+* DOUBLE PRECISION RESID
+* ..
+* .. Array Arguments ..
+* INTEGER IPIV( * )
+* DOUBLE PRECISION A( LDA, * ), AFAC( LDAFAC, * ), C( LDC, * ),
+* $ RWORK( * )
+* ..
+*
+*
+*> \par Purpose:
+* =============
+*>
+*> \verbatim
+*>
+*> DSYT01 reconstructs a symmetric indefinite matrix A from its
+*> block L*D*L' or U*D*U' factorization and computes the residual
+*> norm( C - A ) / ( N * norm(A) * EPS ),
+*> where C is the reconstructed matrix and EPS is the machine epsilon.
+*> \endverbatim
+*
+* Arguments:
+* ==========
+*
+*> \param[in] UPLO
+*> \verbatim
+*> UPLO is CHARACTER*1
+*> Specifies whether the upper or lower triangular part of the
+*> symmetric matrix A is stored:
+*> = 'U': Upper triangular
+*> = 'L': Lower triangular
+*> \endverbatim
+*>
+*> \param[in] N
+*> \verbatim
+*> N is INTEGER
+*> The number of rows and columns of the matrix A. N >= 0.
+*> \endverbatim
+*>
+*> \param[in] A
+*> \verbatim
+*> A is DOUBLE PRECISION array, dimension (LDA,N)
+*> The original symmetric matrix A.
+*> \endverbatim
+*>
+*> \param[in] LDA
+*> \verbatim
+*> LDA is INTEGER
+*> The leading dimension of the array A. LDA >= max(1,N)
+*> \endverbatim
+*>
+*> \param[in] AFAC
+*> \verbatim
+*> AFAC is DOUBLE PRECISION array, dimension (LDAFAC,N)
+*> The factored form of the matrix A. AFAC contains the block
+*> diagonal matrix D and the multipliers used to obtain the
+*> factor L or U from the block L*D*L' or U*D*U' factorization
+*> as computed by DSYTRF.
+*> \endverbatim
+*>
+*> \param[in] LDAFAC
+*> \verbatim
+*> LDAFAC is INTEGER
+*> The leading dimension of the array AFAC. LDAFAC >= max(1,N).
+*> \endverbatim
+*>
+*> \param[in] IPIV
+*> \verbatim
+*> IPIV is INTEGER array, dimension (N)
+*> The pivot indices from DSYTRF.
+*> \endverbatim
+*>
+*> \param[out] C
+*> \verbatim
+*> C is DOUBLE PRECISION array, dimension (LDC,N)
+*> \endverbatim
+*>
+*> \param[in] LDC
+*> \verbatim
+*> LDC is INTEGER
+*> The leading dimension of the array C. LDC >= max(1,N).
+*> \endverbatim
+*>
+*> \param[out] RWORK
+*> \verbatim
+*> RWORK is DOUBLE PRECISION array, dimension (N)
+*> \endverbatim
+*>
+*> \param[out] RESID
+*> \verbatim
+*> RESID is DOUBLE PRECISION
+*> If UPLO = 'L', norm(L*D*L' - A) / ( N * norm(A) * EPS )
+*> If UPLO = 'U', norm(U*D*U' - A) / ( N * norm(A) * EPS )
+*> \endverbatim
+*
+* Authors:
+* ========
+*
+*> \author Univ. of Tennessee
+*> \author Univ. of California Berkeley
+*> \author Univ. of Colorado Denver
+*> \author NAG Ltd.
+*
+*> \date November 2016
+*
+* @precisions fortran d -> s
+*
+*> \ingroup double_lin
+*
+* =====================================================================
+ SUBROUTINE DSYT01_AA( UPLO, N, A, LDA, AFAC, LDAFAC, IPIV, C,
+ $ LDC, RWORK, RESID )
+*
+* -- LAPACK test routine (version 3.5.0) --
+* -- LAPACK is a software package provided by Univ. of Tennessee, --
+* -- Univ. of California Berkeley, Univ. of Colorado Denver and NAG Ltd..--
+* November 2016
+*
+* .. Scalar Arguments ..
+ CHARACTER UPLO
+ INTEGER LDA, LDAFAC, LDC, N
+ DOUBLE PRECISION RESID
+* ..
+* .. Array Arguments ..
+ INTEGER IPIV( * )
+ DOUBLE PRECISION A( LDA, * ), AFAC( LDAFAC, * ), C( LDC, * ),
+ $ RWORK( * )
+* ..
+*
+* =====================================================================
+*
+* .. Parameters ..
+ DOUBLE PRECISION ZERO, ONE
+ PARAMETER ( ZERO = 0.0D+0, ONE = 1.0D+0 )
+* ..
+* .. Local Scalars ..
+ INTEGER I, J
+ DOUBLE PRECISION ANORM, EPS
+* ..
+* .. External Functions ..
+ LOGICAL LSAME
+ DOUBLE PRECISION DLAMCH, DLANSY
+ EXTERNAL LSAME, DLAMCH, DLANSY
+* ..
+* .. External Subroutines ..
+ EXTERNAL DLASET, DLAVSY
+* ..
+* .. Intrinsic Functions ..
+ INTRINSIC DBLE
+* ..
+* .. Executable Statements ..
+*
+* Quick exit if N = 0.
+*
+ IF( N.LE.0 ) THEN
+ RESID = ZERO
+ RETURN
+ END IF
+*
+* Determine EPS and the norm of A.
+*
+ EPS = DLAMCH( 'Epsilon' )
+ ANORM = DLANSY( '1', UPLO, N, A, LDA, RWORK )
+*
+* Initialize C to the tridiagonal matrix T.
+*
+ CALL DLASET( 'Full', N, N, ZERO, ZERO, C, LDC )
+ CALL DLACPY( 'F', 1, N, AFAC( 1, 1 ), LDAFAC+1, C( 1, 1 ), LDC+1 )
+ IF( N.GT.1 ) THEN
+ IF( LSAME( UPLO, 'U' ) ) THEN
+ CALL DLACPY( 'F', 1, N-1, AFAC( 1, 2 ), LDAFAC+1, C( 1, 2 ),
+ $ LDC+1 )
+ CALL DLACPY( 'F', 1, N-1, AFAC( 1, 2 ), LDAFAC+1, C( 2, 1 ),
+ $ LDC+1 )
+ ELSE
+ CALL DLACPY( 'F', 1, N-1, AFAC( 2, 1 ), LDAFAC+1, C( 1, 2 ),
+ $ LDC+1 )
+ CALL DLACPY( 'F', 1, N-1, AFAC( 2, 1 ), LDAFAC+1, C( 2, 1 ),
+ $ LDC+1 )
+ ENDIF
+ ENDIF
+*
+* Call DTRMM to form the product U' * D (or L * D ).
+*
+ IF( LSAME( UPLO, 'U' ) ) THEN
+ CALL DTRMM( 'Left', UPLO, 'Transpose', 'Unit', N-1, N,
+ $ ONE, AFAC( 1, 2 ), LDAFAC, C( 2, 1 ), LDC )
+ ELSE
+ CALL DTRMM( 'Left', UPLO, 'No transpose', 'Unit', N-1, N,
+ $ ONE, AFAC( 2, 1 ), LDAFAC, C( 2, 1 ), LDC )
+ END IF
+*
+* Call DTRMM again to multiply by U (or L ).
+*
+ IF( LSAME( UPLO, 'U' ) ) THEN
+ CALL DTRMM( 'Right', UPLO, 'No transpose', 'Unit', N, N-1,
+ $ ONE, AFAC( 1, 2 ), LDAFAC, C( 1, 2 ), LDC )
+ ELSE
+ CALL DTRMM( 'Right', UPLO, 'Transpose', 'Unit', N, N-1,
+ $ ONE, AFAC( 2, 1 ), LDAFAC, C( 1, 2 ), LDC )
+ END IF
+*
+* Apply symmetric pivots
+*
+ DO J = N, 1, -1
+ I = IPIV( J )
+ IF( I.NE.J )
+ $ CALL DSWAP( N, C( J, 1 ), LDC, C( I, 1 ), LDC )
+ END DO
+ DO J = N, 1, -1
+ I = IPIV( J )
+ IF( I.NE.J )
+ $ CALL DSWAP( N, C( 1, J ), 1, C( 1, I ), 1 )
+ END DO
+*
+*
+* Compute the difference C - A .
+*
+ IF( LSAME( UPLO, 'U' ) ) THEN
+ DO J = 1, N
+ DO I = 1, J
+ C( I, J ) = C( I, J ) - A( I, J )
+ END DO
+ END DO
+ ELSE
+ DO J = 1, N
+ DO I = J, N
+ C( I, J ) = C( I, J ) - A( I, J )
+ END DO
+ END DO
+ END IF
+*
+* Compute norm( C - A ) / ( N * norm(A) * EPS )
+*
+ RESID = DLANSY( '1', UPLO, N, C, LDC, RWORK )
+*
+ IF( ANORM.LE.ZERO ) THEN
+ IF( RESID.NE.ZERO )
+ $ RESID = ONE / EPS
+ ELSE
+ RESID = ( ( RESID / DBLE( N ) ) / ANORM ) / EPS
+ END IF
+*
+ RETURN
+*
+* End of DSYT01
+*
+ END