root / src / lapack / double / dlamrg.f @ 2
Historique | Voir | Annoter | Télécharger (2,8 ko)
| 1 |
SUBROUTINE DLAMRG( N1, N2, A, DTRD1, DTRD2, INDEX ) |
|---|---|
| 2 |
* |
| 3 |
* -- LAPACK 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 DTRD1, DTRD2, N1, N2 |
| 10 |
* .. |
| 11 |
* .. Array Arguments .. |
| 12 |
INTEGER INDEX( * ) |
| 13 |
DOUBLE PRECISION A( * ) |
| 14 |
* .. |
| 15 |
* |
| 16 |
* Purpose |
| 17 |
* ======= |
| 18 |
* |
| 19 |
* DLAMRG will create a permutation list which will merge the elements |
| 20 |
* of A (which is composed of two independently sorted sets) into a |
| 21 |
* single set which is sorted in ascending order. |
| 22 |
* |
| 23 |
* Arguments |
| 24 |
* ========= |
| 25 |
* |
| 26 |
* N1 (input) INTEGER |
| 27 |
* N2 (input) INTEGER |
| 28 |
* These arguements contain the respective lengths of the two |
| 29 |
* sorted lists to be merged. |
| 30 |
* |
| 31 |
* A (input) DOUBLE PRECISION array, dimension (N1+N2) |
| 32 |
* The first N1 elements of A contain a list of numbers which |
| 33 |
* are sorted in either ascending or descending order. Likewise |
| 34 |
* for the final N2 elements. |
| 35 |
* |
| 36 |
* DTRD1 (input) INTEGER |
| 37 |
* DTRD2 (input) INTEGER |
| 38 |
* These are the strides to be taken through the array A. |
| 39 |
* Allowable strides are 1 and -1. They indicate whether a |
| 40 |
* subset of A is sorted in ascending (DTRDx = 1) or descending |
| 41 |
* (DTRDx = -1) order. |
| 42 |
* |
| 43 |
* INDEX (output) INTEGER array, dimension (N1+N2) |
| 44 |
* On exit this array will contain a permutation such that |
| 45 |
* if B( I ) = A( INDEX( I ) ) for I=1,N1+N2, then B will be |
| 46 |
* sorted in ascending order. |
| 47 |
* |
| 48 |
* ===================================================================== |
| 49 |
* |
| 50 |
* .. Local Scalars .. |
| 51 |
INTEGER I, IND1, IND2, N1SV, N2SV |
| 52 |
* .. |
| 53 |
* .. Executable Statements .. |
| 54 |
* |
| 55 |
N1SV = N1 |
| 56 |
N2SV = N2 |
| 57 |
IF( DTRD1.GT.0 ) THEN |
| 58 |
IND1 = 1 |
| 59 |
ELSE |
| 60 |
IND1 = N1 |
| 61 |
END IF |
| 62 |
IF( DTRD2.GT.0 ) THEN |
| 63 |
IND2 = 1 + N1 |
| 64 |
ELSE |
| 65 |
IND2 = N1 + N2 |
| 66 |
END IF |
| 67 |
I = 1 |
| 68 |
* while ( (N1SV > 0) & (N2SV > 0) ) |
| 69 |
10 CONTINUE |
| 70 |
IF( N1SV.GT.0 .AND. N2SV.GT.0 ) THEN |
| 71 |
IF( A( IND1 ).LE.A( IND2 ) ) THEN |
| 72 |
INDEX( I ) = IND1 |
| 73 |
I = I + 1 |
| 74 |
IND1 = IND1 + DTRD1 |
| 75 |
N1SV = N1SV - 1 |
| 76 |
ELSE |
| 77 |
INDEX( I ) = IND2 |
| 78 |
I = I + 1 |
| 79 |
IND2 = IND2 + DTRD2 |
| 80 |
N2SV = N2SV - 1 |
| 81 |
END IF |
| 82 |
GO TO 10 |
| 83 |
END IF |
| 84 |
* end while |
| 85 |
IF( N1SV.EQ.0 ) THEN |
| 86 |
DO 20 N1SV = 1, N2SV |
| 87 |
INDEX( I ) = IND2 |
| 88 |
I = I + 1 |
| 89 |
IND2 = IND2 + DTRD2 |
| 90 |
20 CONTINUE |
| 91 |
ELSE |
| 92 |
* N2SV .EQ. 0 |
| 93 |
DO 30 N2SV = 1, N1SV |
| 94 |
INDEX( I ) = IND1 |
| 95 |
I = I + 1 |
| 96 |
IND1 = IND1 + DTRD1 |
| 97 |
30 CONTINUE |
| 98 |
END IF |
| 99 |
* |
| 100 |
RETURN |
| 101 |
* |
| 102 |
* End of DLAMRG |
| 103 |
* |
| 104 |
END |