[BACK]Return to zporfsx.f CVS log [TXT] [DIR] Up to [local] / rpl / lapack / lapack

Diff for /rpl/lapack/lapack/zporfsx.f between versions 1.14 and 1.15

version 1.14, 2018/05/29 07:18:32 version 1.15, 2020/05/21 21:46:10
Line 44 Line 44
 *> \verbatim  *> \verbatim
 *>  *>
 *>    ZPORFSX improves the computed solution to a system of linear  *>    ZPORFSX improves the computed solution to a system of linear
 *>    equations when the coefficient matrix is symmetric positive  *>    equations when the coefficient matrix is Hermitian positive
 *>    definite, and provides error bounds and backward error estimates  *>    definite, and provides error bounds and backward error estimates
 *>    for the solution.  In addition to normwise error bound, the code  *>    for the solution.  In addition to normwise error bound, the code
 *>    provides maximum componentwise error bound if possible.  See  *>    provides maximum componentwise error bound if possible.  See
Line 103 Line 103
 *> \param[in] A  *> \param[in] A
 *> \verbatim  *> \verbatim
 *>          A is COMPLEX*16 array, dimension (LDA,N)  *>          A is COMPLEX*16 array, dimension (LDA,N)
 *>     The symmetric matrix A.  If UPLO = 'U', the leading N-by-N  *>     The Hermitian matrix A.  If UPLO = 'U', the leading N-by-N
 *>     upper triangular part of A contains the upper triangular part  *>     upper triangular part of A contains the upper triangular part
 *>     of the matrix A, and the strictly lower triangular part of A  *>     of the matrix A, and the strictly lower triangular part of A
 *>     is not referenced.  If UPLO = 'L', the leading N-by-N lower  *>     is not referenced.  If UPLO = 'L', the leading N-by-N lower
Line 134 Line 134
 *> \param[in,out] S  *> \param[in,out] S
 *> \verbatim  *> \verbatim
 *>          S is DOUBLE PRECISION array, dimension (N)  *>          S is DOUBLE PRECISION array, dimension (N)
 *>     The row scale factors for A.  If EQUED = 'Y', A is multiplied on  *>     The scale factors for A.  If EQUED = 'Y', A is multiplied on
 *>     the left and right by diag(S).  S is an input argument if FACT =  *>     the left and right by diag(S).  S is an input argument if FACT =
 *>     'F'; otherwise, S is an output argument.  If FACT = 'F' and EQUED  *>     'F'; otherwise, S is an output argument.  If FACT = 'F' and EQUED
 *>     = 'Y', each element of S must be positive.  If S is output, each  *>     = 'Y', each element of S must be positive.  If S is output, each
Line 262 Line 262
 *>     information as described below. There currently are up to three  *>     information as described below. There currently are up to three
 *>     pieces of information returned for each right-hand side. If  *>     pieces of information returned for each right-hand side. If
 *>     componentwise accuracy is not requested (PARAMS(3) = 0.0), then  *>     componentwise accuracy is not requested (PARAMS(3) = 0.0), then
 *>     ERR_BNDS_COMP is not accessed.  If N_ERR_BNDS .LT. 3, then at most  *>     ERR_BNDS_COMP is not accessed.  If N_ERR_BNDS < 3, then at most
 *>     the first (:,N_ERR_BNDS) entries are returned.  *>     the first (:,N_ERR_BNDS) entries are returned.
 *>  *>
 *>     The first index in ERR_BNDS_COMP(i,:) corresponds to the ith  *>     The first index in ERR_BNDS_COMP(i,:) corresponds to the ith
Line 298 Line 298
 *> \param[in] NPARAMS  *> \param[in] NPARAMS
 *> \verbatim  *> \verbatim
 *>          NPARAMS is INTEGER  *>          NPARAMS is INTEGER
 *>     Specifies the number of parameters set in PARAMS.  If .LE. 0, the  *>     Specifies the number of parameters set in PARAMS.  If <= 0, the
 *>     PARAMS array is never referenced and default values are used.  *>     PARAMS array is never referenced and default values are used.
 *> \endverbatim  *> \endverbatim
 *>  *>
 *> \param[in,out] PARAMS  *> \param[in,out] PARAMS
 *> \verbatim  *> \verbatim
 *>          PARAMS is DOUBLE PRECISION array, dimension NPARAMS  *>          PARAMS is DOUBLE PRECISION array, dimension NPARAMS
 *>     Specifies algorithm parameters.  If an entry is .LT. 0.0, then  *>     Specifies algorithm parameters.  If an entry is < 0.0, then
 *>     that entry will be filled with default value used for that  *>     that entry will be filled with default value used for that
 *>     parameter.  Only positions up to NPARAMS are accessed; defaults  *>     parameter.  Only positions up to NPARAMS are accessed; defaults
 *>     are used for higher-numbered parameters.  *>     are used for higher-numbered parameters.
Line 313 Line 313
 *>       PARAMS(LA_LINRX_ITREF_I = 1) : Whether to perform iterative  *>       PARAMS(LA_LINRX_ITREF_I = 1) : Whether to perform iterative
 *>            refinement or not.  *>            refinement or not.
 *>         Default: 1.0D+0  *>         Default: 1.0D+0
 *>            = 0.0 : No refinement is performed, and no error bounds are  *>            = 0.0:  No refinement is performed, and no error bounds are
 *>                    computed.  *>                    computed.
 *>            = 1.0 : Use the double-precision refinement algorithm,  *>            = 1.0:  Use the double-precision refinement algorithm,
 *>                    possibly with doubled-single computations if the  *>                    possibly with doubled-single computations if the
 *>                    compilation environment does not support DOUBLE  *>                    compilation environment does not support DOUBLE
 *>                    PRECISION.  *>                    PRECISION.
Removed from v.1.14  
changed lines
  Added in v.1.15

CVSweb interface <joel.bertrand@systella.fr>