1: SUBROUTINE DLASWP( N, A, LDA, K1, K2, IPIV, INCX )
2: *
3: * -- LAPACK auxiliary routine (version 3.2) --
4: * -- LAPACK is a software package provided by Univ. of Tennessee, --
5: * -- Univ. of California Berkeley, Univ. of Colorado Denver and NAG Ltd..--
6: * November 2006
7: *
8: * .. Scalar Arguments ..
9: INTEGER INCX, K1, K2, LDA, N
10: * ..
11: * .. Array Arguments ..
12: INTEGER IPIV( * )
13: DOUBLE PRECISION A( LDA, * )
14: * ..
15: *
16: * Purpose
17: * =======
18: *
19: * DLASWP performs a series of row interchanges on the matrix A.
20: * One row interchange is initiated for each of rows K1 through K2 of A.
21: *
22: * Arguments
23: * =========
24: *
25: * N (input) INTEGER
26: * The number of columns of the matrix A.
27: *
28: * A (input/output) DOUBLE PRECISION array, dimension (LDA,N)
29: * On entry, the matrix of column dimension N to which the row
30: * interchanges will be applied.
31: * On exit, the permuted matrix.
32: *
33: * LDA (input) INTEGER
34: * The leading dimension of the array A.
35: *
36: * K1 (input) INTEGER
37: * The first element of IPIV for which a row interchange will
38: * be done.
39: *
40: * K2 (input) INTEGER
41: * The last element of IPIV for which a row interchange will
42: * be done.
43: *
44: * IPIV (input) INTEGER array, dimension (K2*abs(INCX))
45: * The vector of pivot indices. Only the elements in positions
46: * K1 through K2 of IPIV are accessed.
47: * IPIV(K) = L implies rows K and L are to be interchanged.
48: *
49: * INCX (input) INTEGER
50: * The increment between successive values of IPIV. If IPIV
51: * is negative, the pivots are applied in reverse order.
52: *
53: * Further Details
54: * ===============
55: *
56: * Modified by
57: * R. C. Whaley, Computer Science Dept., Univ. of Tenn., Knoxville, USA
58: *
59: * =====================================================================
60: *
61: * .. Local Scalars ..
62: INTEGER I, I1, I2, INC, IP, IX, IX0, J, K, N32
63: DOUBLE PRECISION TEMP
64: * ..
65: * .. Executable Statements ..
66: *
67: * Interchange row I with row IPIV(I) for each of rows K1 through K2.
68: *
69: IF( INCX.GT.0 ) THEN
70: IX0 = K1
71: I1 = K1
72: I2 = K2
73: INC = 1
74: ELSE IF( INCX.LT.0 ) THEN
75: IX0 = 1 + ( 1-K2 )*INCX
76: I1 = K2
77: I2 = K1
78: INC = -1
79: ELSE
80: RETURN
81: END IF
82: *
83: N32 = ( N / 32 )*32
84: IF( N32.NE.0 ) THEN
85: DO 30 J = 1, N32, 32
86: IX = IX0
87: DO 20 I = I1, I2, INC
88: IP = IPIV( IX )
89: IF( IP.NE.I ) THEN
90: DO 10 K = J, J + 31
91: TEMP = A( I, K )
92: A( I, K ) = A( IP, K )
93: A( IP, K ) = TEMP
94: 10 CONTINUE
95: END IF
96: IX = IX + INCX
97: 20 CONTINUE
98: 30 CONTINUE
99: END IF
100: IF( N32.NE.N ) THEN
101: N32 = N32 + 1
102: IX = IX0
103: DO 50 I = I1, I2, INC
104: IP = IPIV( IX )
105: IF( IP.NE.I ) THEN
106: DO 40 K = N32, N
107: TEMP = A( I, K )
108: A( I, K ) = A( IP, K )
109: A( IP, K ) = TEMP
110: 40 CONTINUE
111: END IF
112: IX = IX + INCX
113: 50 CONTINUE
114: END IF
115: *
116: RETURN
117: *
118: * End of DLASWP
119: *
120: END
CVSweb interface <joel.bertrand@systella.fr>