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

1.5     ! bertrand    1: *> \brief \b DGERFSX
        !             2: *
        !             3: *  =========== DOCUMENTATION ===========
        !             4: *
        !             5: * Online html documentation available at 
        !             6: *            http://www.netlib.org/lapack/explore-html/ 
        !             7: *
        !             8: *> \htmlonly
        !             9: *> Download DGERFSX + dependencies 
        !            10: *> <a href="http://www.netlib.org/cgi-bin/netlibfiles.tgz?format=tgz&filename=/lapack/lapack_routine/dgerfsx.f"> 
        !            11: *> [TGZ]</a> 
        !            12: *> <a href="http://www.netlib.org/cgi-bin/netlibfiles.zip?format=zip&filename=/lapack/lapack_routine/dgerfsx.f"> 
        !            13: *> [ZIP]</a> 
        !            14: *> <a href="http://www.netlib.org/cgi-bin/netlibfiles.txt?format=txt&filename=/lapack/lapack_routine/dgerfsx.f"> 
        !            15: *> [TXT]</a>
        !            16: *> \endhtmlonly 
        !            17: *
        !            18: *  Definition:
        !            19: *  ===========
        !            20: *
        !            21: *       SUBROUTINE DGERFSX( TRANS, EQUED, N, NRHS, A, LDA, AF, LDAF, IPIV,
        !            22: *                           R, C, B, LDB, X, LDX, RCOND, BERR, N_ERR_BNDS,
        !            23: *                           ERR_BNDS_NORM, ERR_BNDS_COMP, NPARAMS, PARAMS,
        !            24: *                           WORK, IWORK, INFO )
        !            25: * 
        !            26: *       .. Scalar Arguments ..
        !            27: *       CHARACTER          TRANS, EQUED
        !            28: *       INTEGER            INFO, LDA, LDAF, LDB, LDX, N, NRHS, NPARAMS,
        !            29: *      $                   N_ERR_BNDS
        !            30: *       DOUBLE PRECISION   RCOND
        !            31: *       ..
        !            32: *       .. Array Arguments ..
        !            33: *       INTEGER            IPIV( * ), IWORK( * )
        !            34: *       DOUBLE PRECISION   A( LDA, * ), AF( LDAF, * ), B( LDB, * ),
        !            35: *      $                   X( LDX , * ), WORK( * )
        !            36: *       DOUBLE PRECISION   R( * ), C( * ), PARAMS( * ), BERR( * ),
        !            37: *      $                   ERR_BNDS_NORM( NRHS, * ),
        !            38: *      $                   ERR_BNDS_COMP( NRHS, * )
        !            39: *       ..
        !            40: *  
        !            41: *
        !            42: *> \par Purpose:
        !            43: *  =============
        !            44: *>
        !            45: *> \verbatim
        !            46: *>
        !            47: *>    DGERFSX improves the computed solution to a system of linear
        !            48: *>    equations and provides error bounds and backward error estimates
        !            49: *>    for the solution.  In addition to normwise error bound, the code
        !            50: *>    provides maximum componentwise error bound if possible.  See
        !            51: *>    comments for ERR_BNDS_NORM and ERR_BNDS_COMP for details of the
        !            52: *>    error bounds.
        !            53: *>
        !            54: *>    The original system of linear equations may have been equilibrated
        !            55: *>    before calling this routine, as described by arguments EQUED, R
        !            56: *>    and C below. In this case, the solution and error bounds returned
        !            57: *>    are for the original unequilibrated system.
        !            58: *> \endverbatim
        !            59: *
        !            60: *  Arguments:
        !            61: *  ==========
        !            62: *
        !            63: *> \verbatim
        !            64: *>     Some optional parameters are bundled in the PARAMS array.  These
        !            65: *>     settings determine how refinement is performed, but often the
        !            66: *>     defaults are acceptable.  If the defaults are acceptable, users
        !            67: *>     can pass NPARAMS = 0 which prevents the source code from accessing
        !            68: *>     the PARAMS argument.
        !            69: *> \endverbatim
        !            70: *>
        !            71: *> \param[in] TRANS
        !            72: *> \verbatim
        !            73: *>          TRANS is CHARACTER*1
        !            74: *>     Specifies the form of the system of equations:
        !            75: *>       = 'N':  A * X = B     (No transpose)
        !            76: *>       = 'T':  A**T * X = B  (Transpose)
        !            77: *>       = 'C':  A**H * X = B  (Conjugate transpose = Transpose)
        !            78: *> \endverbatim
        !            79: *>
        !            80: *> \param[in] EQUED
        !            81: *> \verbatim
        !            82: *>          EQUED is CHARACTER*1
        !            83: *>     Specifies the form of equilibration that was done to A
        !            84: *>     before calling this routine. This is needed to compute
        !            85: *>     the solution and error bounds correctly.
        !            86: *>       = 'N':  No equilibration
        !            87: *>       = 'R':  Row equilibration, i.e., A has been premultiplied by
        !            88: *>               diag(R).
        !            89: *>       = 'C':  Column equilibration, i.e., A has been postmultiplied
        !            90: *>               by diag(C).
        !            91: *>       = 'B':  Both row and column equilibration, i.e., A has been
        !            92: *>               replaced by diag(R) * A * diag(C).
        !            93: *>               The right hand side B has been changed accordingly.
        !            94: *> \endverbatim
        !            95: *>
        !            96: *> \param[in] N
        !            97: *> \verbatim
        !            98: *>          N is INTEGER
        !            99: *>     The order of the matrix A.  N >= 0.
        !           100: *> \endverbatim
        !           101: *>
        !           102: *> \param[in] NRHS
        !           103: *> \verbatim
        !           104: *>          NRHS is INTEGER
        !           105: *>     The number of right hand sides, i.e., the number of columns
        !           106: *>     of the matrices B and X.  NRHS >= 0.
        !           107: *> \endverbatim
        !           108: *>
        !           109: *> \param[in] A
        !           110: *> \verbatim
        !           111: *>          A is DOUBLE PRECISION array, dimension (LDA,N)
        !           112: *>     The original N-by-N matrix A.
        !           113: *> \endverbatim
        !           114: *>
        !           115: *> \param[in] LDA
        !           116: *> \verbatim
        !           117: *>          LDA is INTEGER
        !           118: *>     The leading dimension of the array A.  LDA >= max(1,N).
        !           119: *> \endverbatim
        !           120: *>
        !           121: *> \param[in] AF
        !           122: *> \verbatim
        !           123: *>          AF is DOUBLE PRECISION array, dimension (LDAF,N)
        !           124: *>     The factors L and U from the factorization A = P*L*U
        !           125: *>     as computed by DGETRF.
        !           126: *> \endverbatim
        !           127: *>
        !           128: *> \param[in] LDAF
        !           129: *> \verbatim
        !           130: *>          LDAF is INTEGER
        !           131: *>     The leading dimension of the array AF.  LDAF >= max(1,N).
        !           132: *> \endverbatim
        !           133: *>
        !           134: *> \param[in] IPIV
        !           135: *> \verbatim
        !           136: *>          IPIV is INTEGER array, dimension (N)
        !           137: *>     The pivot indices from DGETRF; for 1<=i<=N, row i of the
        !           138: *>     matrix was interchanged with row IPIV(i).
        !           139: *> \endverbatim
        !           140: *>
        !           141: *> \param[in] R
        !           142: *> \verbatim
        !           143: *>          R is DOUBLE PRECISION array, dimension (N)
        !           144: *>     The row scale factors for A.  If EQUED = 'R' or 'B', A is
        !           145: *>     multiplied on the left by diag(R); if EQUED = 'N' or 'C', R
        !           146: *>     is not accessed.  
        !           147: *>     If R is accessed, each element of R should be a power of the radix
        !           148: *>     to ensure a reliable solution and error estimates. Scaling by
        !           149: *>     powers of the radix does not cause rounding errors unless the
        !           150: *>     result underflows or overflows. Rounding errors during scaling
        !           151: *>     lead to refining with a matrix that is not equivalent to the
        !           152: *>     input matrix, producing error estimates that may not be
        !           153: *>     reliable.
        !           154: *> \endverbatim
        !           155: *>
        !           156: *> \param[in] C
        !           157: *> \verbatim
        !           158: *>          C is DOUBLE PRECISION array, dimension (N)
        !           159: *>     The column scale factors for A.  If EQUED = 'C' or 'B', A is
        !           160: *>     multiplied on the right by diag(C); if EQUED = 'N' or 'R', C
        !           161: *>     is not accessed. 
        !           162: *>     If C is accessed, each element of C should be a power of the radix
        !           163: *>     to ensure a reliable solution and error estimates. Scaling by
        !           164: *>     powers of the radix does not cause rounding errors unless the
        !           165: *>     result underflows or overflows. Rounding errors during scaling
        !           166: *>     lead to refining with a matrix that is not equivalent to the
        !           167: *>     input matrix, producing error estimates that may not be
        !           168: *>     reliable.
        !           169: *> \endverbatim
        !           170: *>
        !           171: *> \param[in] B
        !           172: *> \verbatim
        !           173: *>          B is DOUBLE PRECISION array, dimension (LDB,NRHS)
        !           174: *>     The right hand side matrix B.
        !           175: *> \endverbatim
        !           176: *>
        !           177: *> \param[in] LDB
        !           178: *> \verbatim
        !           179: *>          LDB is INTEGER
        !           180: *>     The leading dimension of the array B.  LDB >= max(1,N).
        !           181: *> \endverbatim
        !           182: *>
        !           183: *> \param[in,out] X
        !           184: *> \verbatim
        !           185: *>          X is DOUBLE PRECISION array, dimension (LDX,NRHS)
        !           186: *>     On entry, the solution matrix X, as computed by DGETRS.
        !           187: *>     On exit, the improved solution matrix X.
        !           188: *> \endverbatim
        !           189: *>
        !           190: *> \param[in] LDX
        !           191: *> \verbatim
        !           192: *>          LDX is INTEGER
        !           193: *>     The leading dimension of the array X.  LDX >= max(1,N).
        !           194: *> \endverbatim
        !           195: *>
        !           196: *> \param[out] RCOND
        !           197: *> \verbatim
        !           198: *>          RCOND is DOUBLE PRECISION
        !           199: *>     Reciprocal scaled condition number.  This is an estimate of the
        !           200: *>     reciprocal Skeel condition number of the matrix A after
        !           201: *>     equilibration (if done).  If this is less than the machine
        !           202: *>     precision (in particular, if it is zero), the matrix is singular
        !           203: *>     to working precision.  Note that the error may still be small even
        !           204: *>     if this number is very small and the matrix appears ill-
        !           205: *>     conditioned.
        !           206: *> \endverbatim
        !           207: *>
        !           208: *> \param[out] BERR
        !           209: *> \verbatim
        !           210: *>          BERR is DOUBLE PRECISION array, dimension (NRHS)
        !           211: *>     Componentwise relative backward error.  This is the
        !           212: *>     componentwise relative backward error of each solution vector X(j)
        !           213: *>     (i.e., the smallest relative change in any element of A or B that
        !           214: *>     makes X(j) an exact solution).
        !           215: *> \endverbatim
        !           216: *>
        !           217: *> \param[in] N_ERR_BNDS
        !           218: *> \verbatim
        !           219: *>          N_ERR_BNDS is INTEGER
        !           220: *>     Number of error bounds to return for each right hand side
        !           221: *>     and each type (normwise or componentwise).  See ERR_BNDS_NORM and
        !           222: *>     ERR_BNDS_COMP below.
        !           223: *> \endverbatim
        !           224: *>
        !           225: *> \param[out] ERR_BNDS_NORM
        !           226: *> \verbatim
        !           227: *>          ERR_BNDS_NORM is DOUBLE PRECISION array, dimension (NRHS, N_ERR_BNDS)
        !           228: *>     For each right-hand side, this array contains information about
        !           229: *>     various error bounds and condition numbers corresponding to the
        !           230: *>     normwise relative error, which is defined as follows:
        !           231: *>
        !           232: *>     Normwise relative error in the ith solution vector:
        !           233: *>             max_j (abs(XTRUE(j,i) - X(j,i)))
        !           234: *>            ------------------------------
        !           235: *>                  max_j abs(X(j,i))
        !           236: *>
        !           237: *>     The array is indexed by the type of error information as described
        !           238: *>     below. There currently are up to three pieces of information
        !           239: *>     returned.
        !           240: *>
        !           241: *>     The first index in ERR_BNDS_NORM(i,:) corresponds to the ith
        !           242: *>     right-hand side.
        !           243: *>
        !           244: *>     The second index in ERR_BNDS_NORM(:,err) contains the following
        !           245: *>     three fields:
        !           246: *>     err = 1 "Trust/don't trust" boolean. Trust the answer if the
        !           247: *>              reciprocal condition number is less than the threshold
        !           248: *>              sqrt(n) * dlamch('Epsilon').
        !           249: *>
        !           250: *>     err = 2 "Guaranteed" error bound: The estimated forward error,
        !           251: *>              almost certainly within a factor of 10 of the true error
        !           252: *>              so long as the next entry is greater than the threshold
        !           253: *>              sqrt(n) * dlamch('Epsilon'). This error bound should only
        !           254: *>              be trusted if the previous boolean is true.
        !           255: *>
        !           256: *>     err = 3  Reciprocal condition number: Estimated normwise
        !           257: *>              reciprocal condition number.  Compared with the threshold
        !           258: *>              sqrt(n) * dlamch('Epsilon') to determine if the error
        !           259: *>              estimate is "guaranteed". These reciprocal condition
        !           260: *>              numbers are 1 / (norm(Z^{-1},inf) * norm(Z,inf)) for some
        !           261: *>              appropriately scaled matrix Z.
        !           262: *>              Let Z = S*A, where S scales each row by a power of the
        !           263: *>              radix so all absolute row sums of Z are approximately 1.
        !           264: *>
        !           265: *>     See Lapack Working Note 165 for further details and extra
        !           266: *>     cautions.
        !           267: *> \endverbatim
        !           268: *>
        !           269: *> \param[out] ERR_BNDS_COMP
        !           270: *> \verbatim
        !           271: *>          ERR_BNDS_COMP is DOUBLE PRECISION array, dimension (NRHS, N_ERR_BNDS)
        !           272: *>     For each right-hand side, this array contains information about
        !           273: *>     various error bounds and condition numbers corresponding to the
        !           274: *>     componentwise relative error, which is defined as follows:
        !           275: *>
        !           276: *>     Componentwise relative error in the ith solution vector:
        !           277: *>                    abs(XTRUE(j,i) - X(j,i))
        !           278: *>             max_j ----------------------
        !           279: *>                         abs(X(j,i))
        !           280: *>
        !           281: *>     The array is indexed by the right-hand side i (on which the
        !           282: *>     componentwise relative error depends), and the type of error
        !           283: *>     information as described below. There currently are up to three
        !           284: *>     pieces of information returned for each right-hand side. If
        !           285: *>     componentwise accuracy is not requested (PARAMS(3) = 0.0), then
        !           286: *>     ERR_BNDS_COMP is not accessed.  If N_ERR_BNDS .LT. 3, then at most
        !           287: *>     the first (:,N_ERR_BNDS) entries are returned.
        !           288: *>
        !           289: *>     The first index in ERR_BNDS_COMP(i,:) corresponds to the ith
        !           290: *>     right-hand side.
        !           291: *>
        !           292: *>     The second index in ERR_BNDS_COMP(:,err) contains the following
        !           293: *>     three fields:
        !           294: *>     err = 1 "Trust/don't trust" boolean. Trust the answer if the
        !           295: *>              reciprocal condition number is less than the threshold
        !           296: *>              sqrt(n) * dlamch('Epsilon').
        !           297: *>
        !           298: *>     err = 2 "Guaranteed" error bound: The estimated forward error,
        !           299: *>              almost certainly within a factor of 10 of the true error
        !           300: *>              so long as the next entry is greater than the threshold
        !           301: *>              sqrt(n) * dlamch('Epsilon'). This error bound should only
        !           302: *>              be trusted if the previous boolean is true.
        !           303: *>
        !           304: *>     err = 3  Reciprocal condition number: Estimated componentwise
        !           305: *>              reciprocal condition number.  Compared with the threshold
        !           306: *>              sqrt(n) * dlamch('Epsilon') to determine if the error
        !           307: *>              estimate is "guaranteed". These reciprocal condition
        !           308: *>              numbers are 1 / (norm(Z^{-1},inf) * norm(Z,inf)) for some
        !           309: *>              appropriately scaled matrix Z.
        !           310: *>              Let Z = S*(A*diag(x)), where x is the solution for the
        !           311: *>              current right-hand side and S scales each row of
        !           312: *>              A*diag(x) by a power of the radix so all absolute row
        !           313: *>              sums of Z are approximately 1.
        !           314: *>
        !           315: *>     See Lapack Working Note 165 for further details and extra
        !           316: *>     cautions.
        !           317: *> \endverbatim
        !           318: *>
        !           319: *> \param[in] NPARAMS
        !           320: *> \verbatim
        !           321: *>          NPARAMS is INTEGER
        !           322: *>     Specifies the number of parameters set in PARAMS.  If .LE. 0, the
        !           323: *>     PARAMS array is never referenced and default values are used.
        !           324: *> \endverbatim
        !           325: *>
        !           326: *> \param[in,out] PARAMS
        !           327: *> \verbatim
        !           328: *>          PARAMS is / output) DOUBLE PRECISION array, dimension (NPARAMS)
        !           329: *>     Specifies algorithm parameters.  If an entry is .LT. 0.0, then
        !           330: *>     that entry will be filled with default value used for that
        !           331: *>     parameter.  Only positions up to NPARAMS are accessed; defaults
        !           332: *>     are used for higher-numbered parameters.
        !           333: *>
        !           334: *>       PARAMS(LA_LINRX_ITREF_I = 1) : Whether to perform iterative
        !           335: *>            refinement or not.
        !           336: *>         Default: 1.0D+0
        !           337: *>            = 0.0 : No refinement is performed, and no error bounds are
        !           338: *>                    computed.
        !           339: *>            = 1.0 : Use the double-precision refinement algorithm,
        !           340: *>                    possibly with doubled-single computations if the
        !           341: *>                    compilation environment does not support DOUBLE
        !           342: *>                    PRECISION.
        !           343: *>              (other values are reserved for future use)
        !           344: *>
        !           345: *>       PARAMS(LA_LINRX_ITHRESH_I = 2) : Maximum number of residual
        !           346: *>            computations allowed for refinement.
        !           347: *>         Default: 10
        !           348: *>         Aggressive: Set to 100 to permit convergence using approximate
        !           349: *>                     factorizations or factorizations other than LU. If
        !           350: *>                     the factorization uses a technique other than
        !           351: *>                     Gaussian elimination, the guarantees in
        !           352: *>                     err_bnds_norm and err_bnds_comp may no longer be
        !           353: *>                     trustworthy.
        !           354: *>
        !           355: *>       PARAMS(LA_LINRX_CWISE_I = 3) : Flag determining if the code
        !           356: *>            will attempt to find a solution with small componentwise
        !           357: *>            relative error in the double-precision algorithm.  Positive
        !           358: *>            is true, 0.0 is false.
        !           359: *>         Default: 1.0 (attempt componentwise convergence)
        !           360: *> \endverbatim
        !           361: *>
        !           362: *> \param[out] WORK
        !           363: *> \verbatim
        !           364: *>          WORK is DOUBLE PRECISION array, dimension (4*N)
        !           365: *> \endverbatim
        !           366: *>
        !           367: *> \param[out] IWORK
        !           368: *> \verbatim
        !           369: *>          IWORK is INTEGER array, dimension (N)
        !           370: *> \endverbatim
        !           371: *>
        !           372: *> \param[out] INFO
        !           373: *> \verbatim
        !           374: *>          INFO is INTEGER
        !           375: *>       = 0:  Successful exit. The solution to every right-hand side is
        !           376: *>         guaranteed.
        !           377: *>       < 0:  If INFO = -i, the i-th argument had an illegal value
        !           378: *>       > 0 and <= N:  U(INFO,INFO) is exactly zero.  The factorization
        !           379: *>         has been completed, but the factor U is exactly singular, so
        !           380: *>         the solution and error bounds could not be computed. RCOND = 0
        !           381: *>         is returned.
        !           382: *>       = N+J: The solution corresponding to the Jth right-hand side is
        !           383: *>         not guaranteed. The solutions corresponding to other right-
        !           384: *>         hand sides K with K > J may not be guaranteed as well, but
        !           385: *>         only the first such right-hand side is reported. If a small
        !           386: *>         componentwise error is not requested (PARAMS(3) = 0.0) then
        !           387: *>         the Jth right-hand side is the first with a normwise error
        !           388: *>         bound that is not guaranteed (the smallest J such
        !           389: *>         that ERR_BNDS_NORM(J,1) = 0.0). By default (PARAMS(3) = 1.0)
        !           390: *>         the Jth right-hand side is the first with either a normwise or
        !           391: *>         componentwise error bound that is not guaranteed (the smallest
        !           392: *>         J such that either ERR_BNDS_NORM(J,1) = 0.0 or
        !           393: *>         ERR_BNDS_COMP(J,1) = 0.0). See the definition of
        !           394: *>         ERR_BNDS_NORM(:,1) and ERR_BNDS_COMP(:,1). To get information
        !           395: *>         about all of the right-hand sides check ERR_BNDS_NORM or
        !           396: *>         ERR_BNDS_COMP.
        !           397: *> \endverbatim
        !           398: *
        !           399: *  Authors:
        !           400: *  ========
        !           401: *
        !           402: *> \author Univ. of Tennessee 
        !           403: *> \author Univ. of California Berkeley 
        !           404: *> \author Univ. of Colorado Denver 
        !           405: *> \author NAG Ltd. 
        !           406: *
        !           407: *> \date November 2011
        !           408: *
        !           409: *> \ingroup doubleGEcomputational
        !           410: *
        !           411: *  =====================================================================
1.1       bertrand  412:       SUBROUTINE DGERFSX( TRANS, EQUED, N, NRHS, A, LDA, AF, LDAF, IPIV,
                    413:      $                    R, C, B, LDB, X, LDX, RCOND, BERR, N_ERR_BNDS,
                    414:      $                    ERR_BNDS_NORM, ERR_BNDS_COMP, NPARAMS, PARAMS,
                    415:      $                    WORK, IWORK, INFO )
                    416: *
1.5     ! bertrand  417: *  -- LAPACK computational routine (version 3.4.0) --
        !           418: *  -- LAPACK is a software package provided by Univ. of Tennessee,    --
        !           419: *  -- Univ. of California Berkeley, Univ. of Colorado Denver and NAG Ltd..--
        !           420: *     November 2011
1.1       bertrand  421: *
                    422: *     .. Scalar Arguments ..
                    423:       CHARACTER          TRANS, EQUED
                    424:       INTEGER            INFO, LDA, LDAF, LDB, LDX, N, NRHS, NPARAMS,
                    425:      $                   N_ERR_BNDS
                    426:       DOUBLE PRECISION   RCOND
                    427: *     ..
                    428: *     .. Array Arguments ..
                    429:       INTEGER            IPIV( * ), IWORK( * )
                    430:       DOUBLE PRECISION   A( LDA, * ), AF( LDAF, * ), B( LDB, * ),
                    431:      $                   X( LDX , * ), WORK( * )
                    432:       DOUBLE PRECISION   R( * ), C( * ), PARAMS( * ), BERR( * ),
                    433:      $                   ERR_BNDS_NORM( NRHS, * ),
                    434:      $                   ERR_BNDS_COMP( NRHS, * )
                    435: *     ..
                    436: *
1.5     ! bertrand  437: *  ==================================================================
1.1       bertrand  438: *
                    439: *     .. Parameters ..
                    440:       DOUBLE PRECISION   ZERO, ONE
                    441:       PARAMETER          ( ZERO = 0.0D+0, ONE = 1.0D+0 )
                    442:       DOUBLE PRECISION   ITREF_DEFAULT, ITHRESH_DEFAULT
                    443:       DOUBLE PRECISION   COMPONENTWISE_DEFAULT, RTHRESH_DEFAULT
                    444:       DOUBLE PRECISION   DZTHRESH_DEFAULT
                    445:       PARAMETER          ( ITREF_DEFAULT = 1.0D+0 )
                    446:       PARAMETER          ( ITHRESH_DEFAULT = 10.0D+0 )
                    447:       PARAMETER          ( COMPONENTWISE_DEFAULT = 1.0D+0 )
                    448:       PARAMETER          ( RTHRESH_DEFAULT = 0.5D+0 )
                    449:       PARAMETER          ( DZTHRESH_DEFAULT = 0.25D+0 )
                    450:       INTEGER            LA_LINRX_ITREF_I, LA_LINRX_ITHRESH_I,
                    451:      $                   LA_LINRX_CWISE_I
                    452:       PARAMETER          ( LA_LINRX_ITREF_I = 1,
                    453:      $                   LA_LINRX_ITHRESH_I = 2 )
                    454:       PARAMETER          ( LA_LINRX_CWISE_I = 3 )
                    455:       INTEGER            LA_LINRX_TRUST_I, LA_LINRX_ERR_I,
                    456:      $                   LA_LINRX_RCOND_I
                    457:       PARAMETER          ( LA_LINRX_TRUST_I = 1, LA_LINRX_ERR_I = 2 )
                    458:       PARAMETER          ( LA_LINRX_RCOND_I = 3 )
                    459: *     ..
                    460: *     .. Local Scalars ..
                    461:       CHARACTER(1)       NORM
                    462:       LOGICAL            ROWEQU, COLEQU, NOTRAN
                    463:       INTEGER            J, TRANS_TYPE, PREC_TYPE, REF_TYPE
                    464:       INTEGER            N_NORMS
                    465:       DOUBLE PRECISION   ANORM, RCOND_TMP
                    466:       DOUBLE PRECISION   ILLRCOND_THRESH, ERR_LBND, CWISE_WRONG
                    467:       LOGICAL            IGNORE_CWISE
                    468:       INTEGER            ITHRESH
                    469:       DOUBLE PRECISION   RTHRESH, UNSTABLE_THRESH
                    470: *     ..
                    471: *     .. External Subroutines ..
                    472:       EXTERNAL           XERBLA, DGECON, DLA_GERFSX_EXTENDED
                    473: *     ..
                    474: *     .. Intrinsic Functions ..
                    475:       INTRINSIC          MAX, SQRT
                    476: *     ..
                    477: *     .. External Functions ..
                    478:       EXTERNAL           LSAME, BLAS_FPINFO_X, ILATRANS, ILAPREC
                    479:       EXTERNAL           DLAMCH, DLANGE, DLA_GERCOND
                    480:       DOUBLE PRECISION   DLAMCH, DLANGE, DLA_GERCOND
                    481:       LOGICAL            LSAME
                    482:       INTEGER            BLAS_FPINFO_X
                    483:       INTEGER            ILATRANS, ILAPREC
                    484: *     ..
                    485: *     .. Executable Statements ..
                    486: *
                    487: *     Check the input parameters.
                    488: *
                    489:       INFO = 0
                    490:       TRANS_TYPE = ILATRANS( TRANS )
                    491:       REF_TYPE = INT( ITREF_DEFAULT )
                    492:       IF ( NPARAMS .GE. LA_LINRX_ITREF_I ) THEN
                    493:          IF ( PARAMS( LA_LINRX_ITREF_I ) .LT. 0.0D+0 ) THEN
                    494:             PARAMS( LA_LINRX_ITREF_I ) = ITREF_DEFAULT
                    495:          ELSE
                    496:             REF_TYPE = PARAMS( LA_LINRX_ITREF_I )
                    497:          END IF
                    498:       END IF
                    499: *
                    500: *     Set default parameters.
                    501: *
                    502:       ILLRCOND_THRESH = DBLE( N ) * DLAMCH( 'Epsilon' )
                    503:       ITHRESH = INT( ITHRESH_DEFAULT )
                    504:       RTHRESH = RTHRESH_DEFAULT
                    505:       UNSTABLE_THRESH = DZTHRESH_DEFAULT
                    506:       IGNORE_CWISE = COMPONENTWISE_DEFAULT .EQ. 0.0D+0
                    507: *
                    508:       IF ( NPARAMS.GE.LA_LINRX_ITHRESH_I ) THEN
                    509:          IF ( PARAMS( LA_LINRX_ITHRESH_I ).LT.0.0D+0 ) THEN
                    510:             PARAMS( LA_LINRX_ITHRESH_I ) = ITHRESH
                    511:          ELSE
                    512:             ITHRESH = INT( PARAMS( LA_LINRX_ITHRESH_I ) )
                    513:          END IF
                    514:       END IF
                    515:       IF ( NPARAMS.GE.LA_LINRX_CWISE_I ) THEN
                    516:          IF ( PARAMS( LA_LINRX_CWISE_I ).LT.0.0D+0 ) THEN
                    517:             IF ( IGNORE_CWISE ) THEN
                    518:                PARAMS( LA_LINRX_CWISE_I ) = 0.0D+0
                    519:             ELSE
                    520:                PARAMS( LA_LINRX_CWISE_I ) = 1.0D+0
                    521:             END IF
                    522:          ELSE
                    523:             IGNORE_CWISE = PARAMS( LA_LINRX_CWISE_I ) .EQ. 0.0D+0
                    524:          END IF
                    525:       END IF
                    526:       IF ( REF_TYPE .EQ. 0 .OR. N_ERR_BNDS .EQ. 0 ) THEN
                    527:          N_NORMS = 0
                    528:       ELSE IF ( IGNORE_CWISE ) THEN
                    529:          N_NORMS = 1
                    530:       ELSE
                    531:          N_NORMS = 2
                    532:       END IF
                    533: *
                    534:       NOTRAN = LSAME( TRANS, 'N' )
                    535:       ROWEQU = LSAME( EQUED, 'R' ) .OR. LSAME( EQUED, 'B' )
                    536:       COLEQU = LSAME( EQUED, 'C' ) .OR. LSAME( EQUED, 'B' )
                    537: *
                    538: *     Test input parameters.
                    539: *
                    540:       IF( TRANS_TYPE.EQ.-1 ) THEN
                    541:         INFO = -1
                    542:       ELSE IF( .NOT.ROWEQU .AND. .NOT.COLEQU .AND.
                    543:      $         .NOT.LSAME( EQUED, 'N' ) ) THEN
                    544:         INFO = -2
                    545:       ELSE IF( N.LT.0 ) THEN
                    546:         INFO = -3
                    547:       ELSE IF( NRHS.LT.0 ) THEN
                    548:         INFO = -4
                    549:       ELSE IF( LDA.LT.MAX( 1, N ) ) THEN
                    550:         INFO = -6
                    551:       ELSE IF( LDAF.LT.MAX( 1, N ) ) THEN
                    552:         INFO = -8
                    553:       ELSE IF( LDB.LT.MAX( 1, N ) ) THEN
                    554:         INFO = -13
                    555:       ELSE IF( LDX.LT.MAX( 1, N ) ) THEN
                    556:         INFO = -15
                    557:       END IF
                    558:       IF( INFO.NE.0 ) THEN
                    559:         CALL XERBLA( 'DGERFSX', -INFO )
                    560:         RETURN
                    561:       END IF
                    562: *
                    563: *     Quick return if possible.
                    564: *
                    565:       IF( N.EQ.0 .OR. NRHS.EQ.0 ) THEN
                    566:          RCOND = 1.0D+0
                    567:          DO J = 1, NRHS
                    568:             BERR( J ) = 0.0D+0
                    569:             IF ( N_ERR_BNDS .GE. 1 ) THEN
                    570:                ERR_BNDS_NORM( J, LA_LINRX_TRUST_I) = 1.0D+0
                    571:                ERR_BNDS_COMP( J, LA_LINRX_TRUST_I ) = 1.0D+0
                    572:             END IF
                    573:             IF ( N_ERR_BNDS .GE. 2 ) THEN
                    574:                ERR_BNDS_NORM( J, LA_LINRX_ERR_I) = 0.0D+0
                    575:                ERR_BNDS_COMP( J, LA_LINRX_ERR_I ) = 0.0D+0
                    576:             END IF
                    577:             IF ( N_ERR_BNDS .GE. 3 ) THEN
                    578:                ERR_BNDS_NORM( J, LA_LINRX_RCOND_I) = 1.0D+0
                    579:                ERR_BNDS_COMP( J, LA_LINRX_RCOND_I ) = 1.0D+0
                    580:             END IF
                    581:          END DO
                    582:          RETURN
                    583:       END IF
                    584: *
                    585: *     Default to failure.
                    586: *
                    587:       RCOND = 0.0D+0
                    588:       DO J = 1, NRHS
                    589:          BERR( J ) = 1.0D+0
                    590:          IF ( N_ERR_BNDS .GE. 1 ) THEN
                    591:             ERR_BNDS_NORM( J, LA_LINRX_TRUST_I ) = 1.0D+0
                    592:             ERR_BNDS_COMP( J, LA_LINRX_TRUST_I ) = 1.0D+0
                    593:          END IF
                    594:          IF ( N_ERR_BNDS .GE. 2 ) THEN
                    595:             ERR_BNDS_NORM( J, LA_LINRX_ERR_I ) = 1.0D+0
                    596:             ERR_BNDS_COMP( J, LA_LINRX_ERR_I ) = 1.0D+0
                    597:          END IF
                    598:          IF ( N_ERR_BNDS .GE. 3 ) THEN
                    599:             ERR_BNDS_NORM( J, LA_LINRX_RCOND_I ) = 0.0D+0
                    600:             ERR_BNDS_COMP( J, LA_LINRX_RCOND_I ) = 0.0D+0
                    601:          END IF
                    602:       END DO
                    603: *
                    604: *     Compute the norm of A and the reciprocal of the condition
                    605: *     number of A.
                    606: *
                    607:       IF( NOTRAN ) THEN
                    608:          NORM = 'I'
                    609:       ELSE
                    610:          NORM = '1'
                    611:       END IF
                    612:       ANORM = DLANGE( NORM, N, N, A, LDA, WORK )
                    613:       CALL DGECON( NORM, N, AF, LDAF, ANORM, RCOND, WORK, IWORK, INFO )
                    614: *
                    615: *     Perform refinement on each right-hand side
                    616: *
                    617:       IF ( REF_TYPE .NE. 0 ) THEN
                    618: 
                    619:          PREC_TYPE = ILAPREC( 'E' )
                    620: 
                    621:          IF ( NOTRAN ) THEN
                    622:             CALL DLA_GERFSX_EXTENDED( PREC_TYPE, TRANS_TYPE,  N,
                    623:      $           NRHS, A, LDA, AF, LDAF, IPIV, COLEQU, C, B,
                    624:      $           LDB, X, LDX, BERR, N_NORMS, ERR_BNDS_NORM,
                    625:      $           ERR_BNDS_COMP, WORK(N+1), WORK(1), WORK(2*N+1),
                    626:      $           WORK(1), RCOND, ITHRESH, RTHRESH, UNSTABLE_THRESH,
                    627:      $           IGNORE_CWISE, INFO )
                    628:          ELSE
                    629:             CALL DLA_GERFSX_EXTENDED( PREC_TYPE, TRANS_TYPE,  N,
                    630:      $           NRHS, A, LDA, AF, LDAF, IPIV, ROWEQU, R, B,
                    631:      $           LDB, X, LDX, BERR, N_NORMS, ERR_BNDS_NORM,
                    632:      $           ERR_BNDS_COMP, WORK(N+1), WORK(1), WORK(2*N+1),
                    633:      $           WORK(1), RCOND, ITHRESH, RTHRESH, UNSTABLE_THRESH,
                    634:      $           IGNORE_CWISE, INFO )
                    635:          END IF
                    636:       END IF
                    637: 
                    638:       ERR_LBND = MAX( 10.0D+0, SQRT( DBLE( N ) ) ) * DLAMCH( 'Epsilon' )
                    639:       IF ( N_ERR_BNDS .GE. 1 .AND. N_NORMS .GE. 1 ) THEN
                    640: *
                    641: *     Compute scaled normwise condition number cond(A*C).
                    642: *
                    643:          IF ( COLEQU .AND. NOTRAN ) THEN
                    644:             RCOND_TMP = DLA_GERCOND( TRANS, N, A, LDA, AF, LDAF, IPIV,
                    645:      $           -1, C, INFO, WORK, IWORK )
                    646:          ELSE IF ( ROWEQU .AND. .NOT. NOTRAN ) THEN
                    647:             RCOND_TMP = DLA_GERCOND( TRANS, N, A, LDA, AF, LDAF, IPIV,
                    648:      $           -1, R, INFO, WORK, IWORK )
                    649:          ELSE
                    650:             RCOND_TMP = DLA_GERCOND( TRANS, N, A, LDA, AF, LDAF, IPIV,
                    651:      $           0, R, INFO, WORK, IWORK )
                    652:          END IF
                    653:          DO J = 1, NRHS
                    654: *
                    655: *     Cap the error at 1.0.
                    656: *
                    657:             IF ( N_ERR_BNDS .GE. LA_LINRX_ERR_I
                    658:      $           .AND. ERR_BNDS_NORM( J, LA_LINRX_ERR_I ) .GT. 1.0D+0 )
                    659:      $           ERR_BNDS_NORM( J, LA_LINRX_ERR_I ) = 1.0D+0
                    660: *
                    661: *     Threshold the error (see LAWN).
                    662: *
                    663:             IF ( RCOND_TMP .LT. ILLRCOND_THRESH ) THEN
                    664:                ERR_BNDS_NORM( J, LA_LINRX_ERR_I ) = 1.0D+0
                    665:                ERR_BNDS_NORM( J, LA_LINRX_TRUST_I ) = 0.0D+0
                    666:                IF ( INFO .LE. N ) INFO = N + J
                    667:             ELSE IF ( ERR_BNDS_NORM( J, LA_LINRX_ERR_I ) .LT. ERR_LBND )
                    668:      $     THEN
                    669:                ERR_BNDS_NORM( J, LA_LINRX_ERR_I ) = ERR_LBND
                    670:                ERR_BNDS_NORM( J, LA_LINRX_TRUST_I ) = 1.0D+0
                    671:             END IF
                    672: *
                    673: *     Save the condition number.
                    674: *
                    675:             IF ( N_ERR_BNDS .GE. LA_LINRX_RCOND_I ) THEN
                    676:                ERR_BNDS_NORM( J, LA_LINRX_RCOND_I ) = RCOND_TMP
                    677:             END IF
                    678:          END DO
                    679:       END IF
                    680: 
                    681:       IF ( N_ERR_BNDS .GE. 1 .AND. N_NORMS .GE. 2 ) THEN
                    682: *
                    683: *     Compute componentwise condition number cond(A*diag(Y(:,J))) for
                    684: *     each right-hand side using the current solution as an estimate of
                    685: *     the true solution.  If the componentwise error estimate is too
                    686: *     large, then the solution is a lousy estimate of truth and the
                    687: *     estimated RCOND may be too optimistic.  To avoid misleading users,
                    688: *     the inverse condition number is set to 0.0 when the estimated
                    689: *     cwise error is at least CWISE_WRONG.
                    690: *
                    691:          CWISE_WRONG = SQRT( DLAMCH( 'Epsilon' ) )
                    692:          DO J = 1, NRHS
                    693:             IF ( ERR_BNDS_COMP( J, LA_LINRX_ERR_I ) .LT. CWISE_WRONG )
                    694:      $           THEN
                    695:                RCOND_TMP = DLA_GERCOND( TRANS, N, A, LDA, AF, LDAF,
                    696:      $              IPIV, 1, X(1,J), INFO, WORK, IWORK )
                    697:             ELSE
                    698:                RCOND_TMP = 0.0D+0
                    699:             END IF
                    700: *
                    701: *     Cap the error at 1.0.
                    702: *
                    703:             IF ( N_ERR_BNDS .GE. LA_LINRX_ERR_I
                    704:      $           .AND. ERR_BNDS_COMP( J, LA_LINRX_ERR_I ) .GT. 1.0D+0 )
                    705:      $           ERR_BNDS_COMP( J, LA_LINRX_ERR_I ) = 1.0D+0
                    706: *
                    707: *     Threshold the error (see LAWN).
                    708: *
                    709:             IF ( RCOND_TMP .LT. ILLRCOND_THRESH ) THEN
                    710:                ERR_BNDS_COMP( J, LA_LINRX_ERR_I ) = 1.0D+0
                    711:                ERR_BNDS_COMP( J, LA_LINRX_TRUST_I ) = 0.0D+0
                    712:                IF ( PARAMS( LA_LINRX_CWISE_I ) .EQ. 1.0D+0
                    713:      $              .AND. INFO.LT.N + J ) INFO = N + J
                    714:             ELSE IF ( ERR_BNDS_COMP( J, LA_LINRX_ERR_I )
                    715:      $              .LT. ERR_LBND ) THEN
                    716:                ERR_BNDS_COMP( J, LA_LINRX_ERR_I ) = ERR_LBND
                    717:                ERR_BNDS_COMP( J, LA_LINRX_TRUST_I ) = 1.0D+0
                    718:             END IF
                    719: *
                    720: *     Save the condition number.
                    721: *
                    722:             IF ( N_ERR_BNDS .GE. LA_LINRX_RCOND_I ) THEN
                    723:                ERR_BNDS_COMP( J, LA_LINRX_RCOND_I ) = RCOND_TMP
                    724:             END IF
                    725:          END DO
                    726:       END IF
                    727: *
                    728:       RETURN
                    729: *
                    730: *     End of DGERFSX
                    731: *
                    732:       END

CVSweb interface <joel.bertrand@systella.fr>