| Re: correct way to use INTERPOLATE function [message #55966] |
Fri, 21 September 2007 15:45 |
David Fanning
Messages: 11724
Registered: August 2001
|
Senior Member |
|
|
Ryan. writes:
> I was actually looking over the validity of what I was doing after I
> posted it. I wasn't certain that what I was doing correctly
> represented what I wanted. Thanks for confirming my thoughts. I got
> sidetracked on what I originally wanted to show. I got busy trying to
> debug my code and ended up doing something I shouldn't. I got
> sidetracked with IDL repeatedly crashing out on me because I was
> unintentionally Interpolating a string (Not a good idea, by the way)
> and spent much time on this. Then I got busy trying to get the
> INTERPOLATE or CONGRID functions to work properly. I will rethink my
> original intentions and hopefully stay on track this time. =)
I experience the same problem when I try to write IDL programs
with the EMACS editor. :-)
Cheers,
David
--
David Fanning, Ph.D.
Fanning Software Consulting, Inc.
Coyote's Guide to IDL Programming: http://www.dfanning.com/
Sepore ma de ni thui. ("Perhaps thou speakest truth.")
|
|
|
|
| Re: correct way to use INTERPOLATE function [message #55967 is a reply to message #55966] |
Fri, 21 September 2007 14:59  |
R.G.Stockwell
Messages: 163
Registered: October 2004
|
Senior Member |
|
|
"Ryan." <rchughes@gmail.com> wrote in message
news:1190411412.834849.94160@g4g2000hsf.googlegroups.com...
>> you probably shouldn't be doing that in the first place.
>
> Hi Bob,
>
> I was actually looking over the validity of what I was doing after I
> posted it. I wasn't certain that what I was doing correctly
> represented what I wanted. Thanks for confirming my thoughts. I got
> sidetracked on what I originally wanted to show. I got busy trying to
> debug my code and ended up doing something I shouldn't. I got
> sidetracked with IDL repeatedly crashing out on me because I was
> unintentionally Interpolating a string (Not a good idea, by the way)
> and spent much time on this. Then I got busy trying to get the
> INTERPOLATE or CONGRID functions to work properly. I will rethink my
> original intentions and hopefully stay on track this time. =)
>
> Thanks for pointing it out,
> Ryan.
You are welcome.
I don't mean to overstate the "shouldn't do it" part. Interpolation
is common and usually a fine thing to do.
The part that stood out for me was taking the original array with
10 elements and no x-values related to those 10 elements.
That is fine, because most people use the index number as the
x-coordinate of the data. So you have x = 0,1,2...9.
Then the example is "interpolating" the original data "s" onto an
array "l" that goes from 0 to 49.5. That is outside the bounds of
the original index (the 0,1,..9) so it doesn't make sense.
If you interpolated onto a finer scale than the integer index,
(from 0,1,2...9 to a new x-value array of 0,0.1,0.2,0.3....... 9.0
then that would be fine). That's all I was really getting at.
Cheers,
bob
PS interpolating strings should be able to work
for instance
x = 'liberal'
x2='conservative'
z = interpol([x,x2], 0.5)
print, result
> moderate
sorry, dumb attempt at a joke
|
|
|
|
| Re: correct way to use INTERPOLATE function [message #55968 is a reply to message #55967] |
Fri, 21 September 2007 14:50 |
rchughes
Messages: 26
Registered: April 2006
|
Junior Member |
|
|
> you probably shouldn't be doing that in the first place.
Hi Bob,
I was actually looking over the validity of what I was doing after I
posted it. I wasn't certain that what I was doing correctly
represented what I wanted. Thanks for confirming my thoughts. I got
sidetracked on what I originally wanted to show. I got busy trying to
debug my code and ended up doing something I shouldn't. I got
sidetracked with IDL repeatedly crashing out on me because I was
unintentionally Interpolating a string (Not a good idea, by the way)
and spent much time on this. Then I got busy trying to get the
INTERPOLATE or CONGRID functions to work properly. I will rethink my
original intentions and hopefully stay on track this time. =)
Thanks for pointing it out,
Ryan.
|
|
|
|
| Re: correct way to use INTERPOLATE function [message #55972 is a reply to message #55968] |
Fri, 21 September 2007 14:35 |
pgrigis
Messages: 436
Registered: September 2007
|
Senior Member |
|
|
R.G. Stockwell wrote:
> "Ryan." <rchughes@gmail.com> wrote in message
> news:1190408773.723336.302670@o80g2000hse.googlegroups.com.. .
>>> Is that waht you are trying to do?
>>
>> Thanks Paolo and David. That's not what I wanted to do but it did
>> help me out to produce the plot I want. I want to create a smallarray
>> vs. largearray plot but If I just tried to plot it, IDL will do what
>> it does best and plot only the first N_ELEMENTS(smallarray) points.
>> The code that I did come up with is as follows (It is very close to
>> what Paolo had):
>>
>> l = FINDGEN(100)*0.5
>> s = [10.3, 9.6, 9.2, 8.5, 7.7, 6.9, 5.8, 5.4, 4.7, 4.1]
>> ns = N_ELEMENTS(s)
>> nl = N_ELEMENTS(l)
>>
>> x = FINDGEN(ns)/(ns-1)
>> x2 = FINDGEN(nl)/nl
>>
>> z = INTERPOL(s, x, x2)
>>
>> PLOT, z, l, PSYM=-6
>
>
> Hi Ryan,
>
> I took a quick glance, and for the life of me i can't figure out
> what it is that you really want to do. You are plotting the
> "l" array here as a function of your "z" (which is
> based on your original "s"). I don't think that means
> anything.
Maybe I know what's happening here...
Let's assume you have a low-resolution circle
given by x1 and y1 coordinates:
n=10
t1=findgen(n)/(n-1)*!Pi*2
x1=sin(t1)
y1=cos(t1)
and a high resolution circle given by x2 and y2
n=64
t2=findgen(n)/(n-1)*!Pi*2
x2=sin(t2)
y2=cos(t2)
Then, assume that for some reason you have
access to x1 and to y2 and not to x2, and you
want to try to plot the best possible approximation
of the high-res circle. Then one may need to
do something like:
xx2=interpol(x1,t1,t2)
and plotting (xx2,y2) is an approximation
to the circle (of course, in such a case,
use of the /spline keyword to interpol may
make the plot nicer).
So, this may be something along the line
of what the OP is doing...
Ciao,
Paolo
>
> I assume you have a watered down example of your
> original problem, but it just seems like
> 1) you have it backwards (i.e. you want to plot
>> plot l,z
>
> and 2)
> you probably shouldn't be doing that in the first place.
>
> My point would be that the elements of "s" should have some
> ordinate related to them. Select the proper ordinate to pair with
> the "s"es and just plot that.
>
> Or if you need to interpolate, then interpolate both in the exact
> same way.
>
> Cheers,
> bob
|
|
|
|
| Re: correct way to use INTERPOLATE function [message #55973 is a reply to message #55972] |
Fri, 21 September 2007 14:16 |
R.G.Stockwell
Messages: 163
Registered: October 2004
|
Senior Member |
|
|
"Ryan." <rchughes@gmail.com> wrote in message
news:1190408773.723336.302670@o80g2000hse.googlegroups.com.. .
>> Is that waht you are trying to do?
>
> Thanks Paolo and David. That's not what I wanted to do but it did
> help me out to produce the plot I want. I want to create a smallarray
> vs. largearray plot but If I just tried to plot it, IDL will do what
> it does best and plot only the first N_ELEMENTS(smallarray) points.
> The code that I did come up with is as follows (It is very close to
> what Paolo had):
>
> l = FINDGEN(100)*0.5
> s = [10.3, 9.6, 9.2, 8.5, 7.7, 6.9, 5.8, 5.4, 4.7, 4.1]
> ns = N_ELEMENTS(s)
> nl = N_ELEMENTS(l)
>
> x = FINDGEN(ns)/(ns-1)
> x2 = FINDGEN(nl)/nl
>
> z = INTERPOL(s, x, x2)
>
> PLOT, z, l, PSYM=-6
Hi Ryan,
I took a quick glance, and for the life of me i can't figure out
what it is that you really want to do. You are plotting the
"l" array here as a function of your "z" (which is
based on your original "s"). I don't think that means
anything.
I assume you have a watered down example of your
original problem, but it just seems like
1) you have it backwards (i.e. you want to plot
> plot l,z
and 2)
you probably shouldn't be doing that in the first place.
My point would be that the elements of "s" should have some
ordinate related to them. Select the proper ordinate to pair with
the "s"es and just plot that.
Or if you need to interpolate, then interpolate both in the exact
same way.
Cheers,
bob
|
|
|
|
| Re: correct way to use INTERPOLATE function [message #55974 is a reply to message #55973] |
Fri, 21 September 2007 14:06 |
rchughes
Messages: 26
Registered: April 2006
|
Junior Member |
|
|
> Is that waht you are trying to do?
Thanks Paolo and David. That's not what I wanted to do but it did
help me out to produce the plot I want. I want to create a smallarray
vs. largearray plot but If I just tried to plot it, IDL will do what
it does best and plot only the first N_ELEMENTS(smallarray) points.
The code that I did come up with is as follows (It is very close to
what Paolo had):
l = FINDGEN(100)*0.5
s = [10.3, 9.6, 9.2, 8.5, 7.7, 6.9, 5.8, 5.4, 4.7, 4.1]
ns = N_ELEMENTS(s)
nl = N_ELEMENTS(l)
x = FINDGEN(ns)/(ns-1)
x2 = FINDGEN(nl)/nl
z = INTERPOL(s, x, x2)
PLOT, z, l, PSYM=-6
Thanks for your help,
Ryan.
|
|
|
|
| Re: correct way to use INTERPOLATE function [message #55976 is a reply to message #55974] |
Fri, 21 September 2007 13:49 |
David Fanning
Messages: 11724
Registered: August 2001
|
Senior Member |
|
|
pgrigis@gmail.com writes:
> For that purpose, you can also use INTERPOL (rather than INTERPOLATE).
> Here's an example:
>
> x=findgen(6)/5
> y=[0.,2,2,5,6,10]
>
> ;first data set
> plot,x,y,psym=-6
>
> x2=findgen(100)/100
> z=interpol(y,x,x2)
> ;the z array is interpolated linearly from the point in y
>
> oplot,x2,z,psym=-4
>
> Is that waht you are trying to do?
Ah, interesting. After some playing around with CONGRID
keywords, I discovered that this interpolation method
is virtually identical to this:
nelem = N_Elements(x2)
OPLOT, x2, CONGRID(y, nelem, /INTERP, /MINUS_ONE)
Could be an interesting article in here somewhere. :-)
Cheers,
David
--
David Fanning, Ph.D.
Fanning Software Consulting, Inc.
Coyote's Guide to IDL Programming: http://www.dfanning.com/
Sepore ma de ni thui. ("Perhaps thou speakest truth.")
|
|
|
|
| Re: correct way to use INTERPOLATE function [message #55977 is a reply to message #55976] |
Fri, 21 September 2007 13:26 |
pgrigis
Messages: 436
Registered: September 2007
|
Senior Member |
|
|
Ryan. wrote:
>> PLOT, Congrid(smallarray, N_Elements(largearray)), largearray
>
> If I do that I get the same result:
> PLOT, CONGRID(smallarray, N_ELEMENTS(largearray), 1, 1) <- With our
> without the INTERP keyword
>
> I want to *stretch* the small array so that it has the same number of
> elements as the large one so I can plot it and all the new points to
> be interpolated linearly between each point. Evenly spaced points is
> fine for my purposes.
For that purpose, you can also use INTERPOL (rather than INTERPOLATE).
Here's an example:
x=findgen(6)/5
y=[0.,2,2,5,6,10]
;first data set
plot,x,y,psym=-6
x2=findgen(100)/100
z=interpol(y,x,x2)
;the z array is interpolated linearly from the point in y
oplot,x2,z,psym=-4
Is that waht you are trying to do?
Cheers,
Paolo
PS: switching to google groups as it turns out newsgroup server
are not supported at my new institution... at least the RSS feed
can be shown in thunderbird in such a way as to look kind of
similar as it should when using a real newsgroup server... but
I am wondering whether there is a way of using a fixed size font
in the google groups "post message" interface...
>
> Ryan.
|
|
|
|
| Re: correct way to use INTERPOLATE function [message #55978 is a reply to message #55977] |
Fri, 21 September 2007 12:51 |
David Fanning
Messages: 11724
Registered: August 2001
|
Senior Member |
|
|
Ryan. writes:
> I'm aware of how to set it. I guess my comment wasn't too clear. I
> meant to say that I get the same result whether I use the INTERP
> keyword or not.
>
> Does your statement work as it is?
No, this is pseudocode. :-)
Try this:
s = [10.3, 9.6, 9.2, 8.5, 7.7, 6.9, 5.8, 5.4, 4.7, 4.1]
l = Findgen(100) * 0.5
P.MULTI=[0,1,2]
Window
Plot, Congrid(s, N_Elements(l))
Plot, Congrid(s, N_Elements(l), /INTERP)
!P.MULTI=0
Cheers,
David
--
David Fanning, Ph.D.
Fanning Software Consulting, Inc.
Coyote's Guide to IDL Programming: http://www.dfanning.com/
Sepore ma de ni thui. ("Perhaps thou speakest truth.")
|
|
|
|
| Re: correct way to use INTERPOLATE function [message #55979 is a reply to message #55978] |
Fri, 21 September 2007 12:37 |
rchughes
Messages: 26
Registered: April 2006
|
Junior Member |
|
|
> The INTERP keyword will be set like this:
>
> Plot, Congrid(sm, N_Elements(large), /INTERP)
>
> Or, like this:
>
> Plot, Congrid(sm, N_Elements(large), INTERP=1)
I'm aware of how to set it. I guess my comment wasn't too clear. I
meant to say that I get the same result whether I use the INTERP
keyword or not.
Does your statement work as it is? If I use the statement like you
wrote it, it asks me for additional arguments (Error: % Variable is
undefined: Y. % Error occurred at: CONGRID 135). Therefore I need to
add the Y and Z arguments which I just put values of 1.
Ryan.
|
|
|
|
| Re: correct way to use INTERPOLATE function [message #55980 is a reply to message #55979] |
Fri, 21 September 2007 12:30 |
David Fanning
Messages: 11724
Registered: August 2001
|
Senior Member |
|
|
Ryan. writes:
> If I do that I get the same result:
> PLOT, CONGRID(smallarray, N_ELEMENTS(largearray), 1, 1) <- With our
> without the INTERP keyword
The INTERP keyword will be set like this:
Plot, Congrid(sm, N_Elements(large), /INTERP)
Or, like this:
Plot, Congrid(sm, N_Elements(large), INTERP=1)
Cheers,
David
--
David Fanning, Ph.D.
Fanning Software Consulting, Inc.
Coyote's Guide to IDL Programming: http://www.dfanning.com/
Sepore ma de ni thui. ("Perhaps thou speakest truth.")
|
|
|
|
| Re: correct way to use INTERPOLATE function [message #55981 is a reply to message #55980] |
Fri, 21 September 2007 12:24 |
rchughes
Messages: 26
Registered: April 2006
|
Junior Member |
|
|
> PLOT, Congrid(smallarray, N_Elements(largearray)), largearray
If I do that I get the same result:
PLOT, CONGRID(smallarray, N_ELEMENTS(largearray), 1, 1) <- With our
without the INTERP keyword
I want to *stretch* the small array so that it has the same number of
elements as the large one so I can plot it and all the new points to
be interpolated linearly between each point. Evenly spaced points is
fine for my purposes.
Ryan.
|
|
|
|
| Re: correct way to use INTERPOLATE function [message #55982 is a reply to message #55981] |
Fri, 21 September 2007 11:42 |
David Fanning
Messages: 11724
Registered: August 2001
|
Senior Member |
|
|
Ryan. writes:
> I want to plot two arrays of different sizes so I am trying to use
> INTERPOLATE on the smaller array to make it the same size as the large
> one. I can't seem to use the INTERPOLATE function correctly and there
> isn't very much documentation on how to generalize it's use. I need
> help in generalizing its use. All I get as a result is the first
> element repeated (the same result I would get if I used the REPLICATE
> function). I want to do this in the general case because I need to do
> it a few times. Here is a simpler version of what I am doing:
>
> smallarray = [10.3, 9.6, 9.2, 8.5, 7.7, 6.9, 5.8, 5.4, 4.7, 4.1]
> largearray = FINDGEN(1000)*0.5
>
> smsize = N_ELEMENTS(smallarray)
> lasize = N_ELEMENTS(largearray)
>
> interpped = INTERPOLATE(smallarray, smsize/(lasize-1)*FINDGEN(lasize))
>
> PLOT, interpped, largearray
>
> One feature of the small array is that it is decreasing throughout and
> has approximately 100 elements. The large array has several thousand
> elements. I believe there is an error in the equation I have in the
> second argument.
It's not clear to me exactly what you are hoping to do,
but it seems clear INTERPOLATE isn't the way to do it. :-)
I think something like this will work:
PLOT, Congrid(smallarray, N_Elements(largearray)), largearray
This will not change the values in smallarray that get plotted.
If you want to interpolate the values, set the INTERP keyword
to CONGRID.
Cheers,
David
--
David Fanning, Ph.D.
Fanning Software Consulting, Inc.
Coyote's Guide to IDL Programming: http://www.dfanning.com/
Sepore ma de ni thui. ("Perhaps thou speakest truth.")
|
|
|
|
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
