| Distances in 3D [message #46003] |
Thu, 27 October 2005 12:17  |
panblosky
Messages: 17
Registered: May 2005
|
Junior Member |
|
|
Hi all,
probably this is a knwon question with a known solution, but I
haven't figure it out yet. I have points in 3D, and I want to calculate
the distances between them, but not the distance between the same
point. This is because I want to know the total sum of the inverse of
those distances. The first thing it came to my mind was to do it with a
for loop, but of course, for large number of points (which is the
case), takes for ever (basically a nxn operation). This is the code I
did (very simple, brutal way):
; xp,yp,zp are the positions
xold=xp
yold=yp
zold=zp
np=n_elements(xp)
dist=fltarr(np)
for i=0,np-1 do begin
xp[i]=0.
esc1=where(xp ne 0,cc1) ;removing the i-th particle
newxp=xp[esc1]
yp[i]=0.
esc2=where(yp ne 0,cc2)
newyp=yp[esc2]
zp[i]=0.
esc3=where(zp ne 0,cc3)
newzp=zp[esc3]
dist[i]=total(1./sqrt((newxp-xold(i))^2.+(newyp-yold(i))^2.+ (newzp-zold(i))^2.))
xp[i]=xold[i]
yp[i]=yold[i] ; restoring the i-th particle
zp[i]=zold[i]
endfor
Any idea of how to make this better? Thanks a lot!!
Andres
|
|
|
|
|
|
| Re: Distances in 3D [message #46142 is a reply to message #46003] |
Thu, 27 October 2005 18:35  |
biophys
Messages: 68
Registered: July 2004
|
Member |
|
|
well, I don't see the point of using 3 where functions while they just
do exactly the same thing here. basically esc1=esc2=esc3, so you only
need 1 where() per loop. i guess shift function should work faster even
if you correct that redundant where functions. here's a one 1d example
for your reference.
x=randomu(systime(1),100)
y=x
t0=systime(1)
for i=0,99 do begin
y[i]=total(1/abs(x[1:99]-x[0]))
x=shift(x,-1)
endfor
print,systime(1)-t0
|
|
|
|
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
