8.3.3 pvcopy

La función "PyPBLAS.pvcopy" distribuye (si es necesario) y copia el contenido de un vector x en otro vector y. Se ha tener en cuenta que los parámetros de entrada y salida son de tipo PyACTS. Los mínimos parametros de entrada de la función son:

from PyACTS import *
import PyACTS.PyPBLAS as PyPBLAS
from RandomArray import *
from Numeric import *
#Dimension of Arrays
n=8
#Initiliaze the Grid
PyACTS.gridinit()
if PyACTS.iread==1:
        print "Example of using PyPBLAS: PvCOPY"
        print "N=",n,";nprow x npcol:",PyACTS.nprow,"x",PyACTS.npcol
        print "Block's size:",PyACTS.mb,"*",PyACTS.nb
        x=reshape(array(range(n),Float),[n,1])
        print "x=",transpose(x)
else:
        x=None
#We convert Numeric Array to PyACTS.Scalapack Array
ACTS_lib=1 # 1=Scalapack
x=Num2PyACTS(x,ACTS_lib)
#We call PBLAS routine
y= PyPBLAS.pvcopy(x)
if PyACTS.ictxt<>-1:
        print "x[",PyACTS.myrow,",",PyACTS.mycol,"]=",transpose(x.data)
        print "y[",PyACTS.myrow,",",PyACTS.mycol,"]=",transpose(y.data)
PyACTS.gridexit()

[vgaliano@localhost EXAMPLES]$ mpirun -np 4 mpipython exPypvcopy.py
Example of using PyPBLAS: PvCOPY
N= 8 ;nprow x npcol: 2 x 2
Block's size: 2 * 2
x= [ [ 0.  1.  2.  3.  4.  5.  6.  7.]]
x[ 0 , 0 ]= [ 0.  1.  4.  5.]
y[ 0 , 0 ]= [ [ 0.  1.  4.  5.]]
x[ 0 , 1 ]= zeros((0,), 'd')
y[ 0 , 1 ]= zeros((0, 4), 'f')
x[ 1 , 0 ]= [ 2.  3.  6.  7.]
y[ 1 , 0 ]= [ [ 2.  3.  6.  7.]]
x[ 1 , 1 ]= zeros((0,), 'd')
y[ 1 , 1 ]= zeros((0, 4), 'f')