Centre Blaise Pascal » HPL sur GPU

.TH HPL_perm 3 "September 10, 2008" "HPL 2.0" "HPL Library Functions"

.SH NAME

HPL_perm \- Combine 2 index arrays - Generate the permutation.

.SH SYNOPSIS

\fB\&#include "hpl.h"\fR

\fB\&void\fR

\fB\&HPL_perm(\fR

\fB\&const int\fR

\fI\&N\fR,

\fB\&int *\fR

\fI\&LINDXA\fR,

\fB\&int *\fR

\fI\&LINDXAU\fR,

\fB\&int *\fR

\fI\&IWORK\fR

\fB\&);\fR

.SH DESCRIPTION

\fB\&HPL_perm\fR

combines  two  index  arrays  and generate the corresponding

permutation. First, this function computes the inverse of LINDXA, and

then combine it with LINDXAU.  Second, in order to be able to perform

the permutation in place,  LINDXAU  is overwritten by the sequence of

permutation  producing  the  same result.  What we ultimately want to

achieve is:  U[LINDXAU[i]] := U[LINDXA[i]] for i in [0..N). After the

call to this function,  this in place permutation can be performed by

for i in [0..N) swap U[i] with U[LINDXAU[i]].

.SH ARGUMENTS

.TP 8

N       (global input)          const int

On entry,  N  specifies the length of the arrays  LINDXA  and

LINDXAU. N should be at least zero.

.TP 8

LINDXA  (global input/output)   int *

On entry,  LINDXA  is an array of dimension N  containing the

source indexes. On exit,  LINDXA  contains the combined index

array.

.TP 8

LINDXAU (global input/output)   int *

On entry,  LINDXAU is an array of dimension N  containing the

target indexes.  On exit,  LINDXAU  contains  the sequence of

permutation,  that  should be applied  in increasing order to

permute the underlying array U in place.

.TP 8

IWORK   (workspace)             int *

On entry, IWORK is a workarray of dimension N.

.SH SEE ALSO

.BR HPL_plindx1 \ (3),

.BR HPL_pdlaswp01N \ (3),

.BR HPL_pdlaswp01T \ (3).