Annotation of rpl/lapack/lapack/dlaswp.f, revision 1.17

1.11      bertrand    1: *> \brief \b DLASWP performs a series of row interchanges on a general rectangular matrix.
1.8       bertrand    2: *
                      3: *  =========== DOCUMENTATION ===========
                      4: *
1.15      bertrand    5: * Online html documentation available at
                      6: *            http://www.netlib.org/lapack/explore-html/
1.8       bertrand    7: *
                      8: *> \htmlonly
1.15      bertrand    9: *> Download DLASWP + dependencies
                     10: *> <a href="http://www.netlib.org/cgi-bin/netlibfiles.tgz?format=tgz&filename=/lapack/lapack_routine/dlaswp.f">
                     11: *> [TGZ]</a>
                     12: *> <a href="http://www.netlib.org/cgi-bin/netlibfiles.zip?format=zip&filename=/lapack/lapack_routine/dlaswp.f">
                     13: *> [ZIP]</a>
                     14: *> <a href="http://www.netlib.org/cgi-bin/netlibfiles.txt?format=txt&filename=/lapack/lapack_routine/dlaswp.f">
1.8       bertrand   15: *> [TXT]</a>
1.15      bertrand   16: *> \endhtmlonly
1.8       bertrand   17: *
                     18: *  Definition:
                     19: *  ===========
                     20: *
                     21: *       SUBROUTINE DLASWP( N, A, LDA, K1, K2, IPIV, INCX )
1.15      bertrand   22: *
1.8       bertrand   23: *       .. Scalar Arguments ..
                     24: *       INTEGER            INCX, K1, K2, LDA, N
                     25: *       ..
                     26: *       .. Array Arguments ..
                     27: *       INTEGER            IPIV( * )
                     28: *       DOUBLE PRECISION   A( LDA, * )
                     29: *       ..
1.15      bertrand   30: *
1.8       bertrand   31: *
                     32: *> \par Purpose:
                     33: *  =============
                     34: *>
                     35: *> \verbatim
                     36: *>
                     37: *> DLASWP performs a series of row interchanges on the matrix A.
                     38: *> One row interchange is initiated for each of rows K1 through K2 of A.
                     39: *> \endverbatim
                     40: *
                     41: *  Arguments:
                     42: *  ==========
                     43: *
                     44: *> \param[in] N
                     45: *> \verbatim
                     46: *>          N is INTEGER
                     47: *>          The number of columns of the matrix A.
                     48: *> \endverbatim
                     49: *>
                     50: *> \param[in,out] A
                     51: *> \verbatim
                     52: *>          A is DOUBLE PRECISION array, dimension (LDA,N)
                     53: *>          On entry, the matrix of column dimension N to which the row
                     54: *>          interchanges will be applied.
                     55: *>          On exit, the permuted matrix.
                     56: *> \endverbatim
                     57: *>
                     58: *> \param[in] LDA
                     59: *> \verbatim
                     60: *>          LDA is INTEGER
                     61: *>          The leading dimension of the array A.
                     62: *> \endverbatim
                     63: *>
                     64: *> \param[in] K1
                     65: *> \verbatim
                     66: *>          K1 is INTEGER
                     67: *>          The first element of IPIV for which a row interchange will
                     68: *>          be done.
                     69: *> \endverbatim
                     70: *>
                     71: *> \param[in] K2
                     72: *> \verbatim
                     73: *>          K2 is INTEGER
1.15      bertrand   74: *>          (K2-K1+1) is the number of elements of IPIV for which a row
                     75: *>          interchange will be done.
1.8       bertrand   76: *> \endverbatim
                     77: *>
                     78: *> \param[in] IPIV
                     79: *> \verbatim
1.15      bertrand   80: *>          IPIV is INTEGER array, dimension (K1+(K2-K1)*abs(INCX))
                     81: *>          The vector of pivot indices. Only the elements in positions
1.17    ! bertrand   82: *>          K1 through K1+(K2-K1)*abs(INCX) of IPIV are accessed.
        !            83: *>          IPIV(K1+(K-K1)*abs(INCX)) = L implies rows K and L are to be
        !            84: *>          interchanged.
1.8       bertrand   85: *> \endverbatim
                     86: *>
                     87: *> \param[in] INCX
                     88: *> \verbatim
                     89: *>          INCX is INTEGER
1.17    ! bertrand   90: *>          The increment between successive values of IPIV. If INCX
1.8       bertrand   91: *>          is negative, the pivots are applied in reverse order.
                     92: *> \endverbatim
                     93: *
                     94: *  Authors:
                     95: *  ========
                     96: *
1.15      bertrand   97: *> \author Univ. of Tennessee
                     98: *> \author Univ. of California Berkeley
                     99: *> \author Univ. of Colorado Denver
                    100: *> \author NAG Ltd.
1.8       bertrand  101: *
1.17    ! bertrand  102: *> \date June 2017
1.8       bertrand  103: *
                    104: *> \ingroup doubleOTHERauxiliary
                    105: *
                    106: *> \par Further Details:
                    107: *  =====================
                    108: *>
                    109: *> \verbatim
                    110: *>
                    111: *>  Modified by
                    112: *>   R. C. Whaley, Computer Science Dept., Univ. of Tenn., Knoxville, USA
                    113: *> \endverbatim
                    114: *>
                    115: *  =====================================================================
1.1       bertrand  116:       SUBROUTINE DLASWP( N, A, LDA, K1, K2, IPIV, INCX )
                    117: *
1.17    ! bertrand  118: *  -- LAPACK auxiliary routine (version 3.7.1) --
1.1       bertrand  119: *  -- LAPACK is a software package provided by Univ. of Tennessee,    --
                    120: *  -- Univ. of California Berkeley, Univ. of Colorado Denver and NAG Ltd..--
1.17    ! bertrand  121: *     June 2017
1.1       bertrand  122: *
                    123: *     .. Scalar Arguments ..
                    124:       INTEGER            INCX, K1, K2, LDA, N
                    125: *     ..
                    126: *     .. Array Arguments ..
                    127:       INTEGER            IPIV( * )
                    128:       DOUBLE PRECISION   A( LDA, * )
                    129: *     ..
                    130: *
                    131: * =====================================================================
                    132: *
                    133: *     .. Local Scalars ..
                    134:       INTEGER            I, I1, I2, INC, IP, IX, IX0, J, K, N32
                    135:       DOUBLE PRECISION   TEMP
                    136: *     ..
                    137: *     .. Executable Statements ..
                    138: *
1.17    ! bertrand  139: *     Interchange row I with row IPIV(K1+(I-K1)*abs(INCX)) for each of rows
        !           140: *     K1 through K2.
1.1       bertrand  141: *
                    142:       IF( INCX.GT.0 ) THEN
                    143:          IX0 = K1
                    144:          I1 = K1
                    145:          I2 = K2
                    146:          INC = 1
                    147:       ELSE IF( INCX.LT.0 ) THEN
1.15      bertrand  148:          IX0 = K1 + ( K1-K2 )*INCX
1.1       bertrand  149:          I1 = K2
                    150:          I2 = K1
                    151:          INC = -1
                    152:       ELSE
                    153:          RETURN
                    154:       END IF
                    155: *
                    156:       N32 = ( N / 32 )*32
                    157:       IF( N32.NE.0 ) THEN
                    158:          DO 30 J = 1, N32, 32
                    159:             IX = IX0
                    160:             DO 20 I = I1, I2, INC
                    161:                IP = IPIV( IX )
                    162:                IF( IP.NE.I ) THEN
                    163:                   DO 10 K = J, J + 31
                    164:                      TEMP = A( I, K )
                    165:                      A( I, K ) = A( IP, K )
                    166:                      A( IP, K ) = TEMP
                    167:    10             CONTINUE
                    168:                END IF
                    169:                IX = IX + INCX
                    170:    20       CONTINUE
                    171:    30    CONTINUE
                    172:       END IF
                    173:       IF( N32.NE.N ) THEN
                    174:          N32 = N32 + 1
                    175:          IX = IX0
                    176:          DO 50 I = I1, I2, INC
                    177:             IP = IPIV( IX )
                    178:             IF( IP.NE.I ) THEN
                    179:                DO 40 K = N32, N
                    180:                   TEMP = A( I, K )
                    181:                   A( I, K ) = A( IP, K )
                    182:                   A( IP, K ) = TEMP
                    183:    40          CONTINUE
                    184:             END IF
                    185:             IX = IX + INCX
                    186:    50    CONTINUE
                    187:       END IF
                    188: *
                    189:       RETURN
                    190: *
                    191: *     End of DLASWP
                    192: *
                    193:       END

CVSweb interface <joel.bertrand@systella.fr>