| SPEED UP [message #7153] |
Thu, 10 October 1996 00:00 |
Mario Noyon
Messages: 11
Registered: August 1996
|
Junior Member |
|
|
Anyone knows that FOR staements slows down IDL.
Well, I would like to present you two cases where I would enjoy to avoid
them and I didn't find any solution yet. If someone could help me :
1 - I want to open a cut a big data of dimension N in n little sets of
dimension n/N. Msut I use a for statement or is there another one
solution?
2 - I have posted this news before but as I didn't get any solution, I
post it once more (because before every body was in holiday). I want to
make a linear fit or an integration with some matrix and only
IDL'routines give me a sufiscient acuracy.
My problem is :
x=ARRAY(n,n,p)
y=ARRAY(p)
and I must do :
for i=0,n-1 do
for j=0,n-1 do
coef(i,j)=INT_TABULATED(x(i,j,*),y)
end
end
And that takes too much time !!!
It is the same for the linear regression. I watched the code of
INT_TABULATED to modify it so it can take arrays of (n,n,p) instead of
arrays of (p). But I couldn't do anything(nor for POLY_FIT)...
If someone has an idea...
PS: Does anyone know how to make in an efficient way the identity
matrix.
Thanks.
--
NOYON Mario
Computer Science in Medical imaging
University of Bordeaux 2
mnoyon@jmc-luni.u-bordeaux2.fr
|
|
|
|
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
