| I need a bit of help....Convol and functions [message #50455] |
Sun, 01 October 2006 03:42  |
D.Kochman@gmail.com
Messages: 7
Registered: October 2006
|
Junior Member |
|
|
So, I'm fairly new to IDL (and an organic chemist so programming is not
my forte), but I'm chunking my way through it. I'm currently in the
process of modifying a program that fits exponential decays given the
impulse response function and the decay curve.
I have to remodel it to fit another much more complex function than an
exponential decay, however with a similar number of parameters. I've
already fixed the GUI, and changed all the references to the widgets,
along with adjusting the appropriate arrays. It now compiles after
many hours of debugging and displays itself appropriately with a dummy
function with the appropriate amount of parameters.
I'm stuck with now implementing the function itself, any help on the
implementation will be *highly* appreciated.
The function is an infinite sum convolved with an exponential decay.
I've done modeling with the sum, and it converges fairly rapidly, and I
can limit it to 10 terms or so and still get accuracy to 6 decimal
places.
Approximately it is:
Sum[(-1)^n*cos(n*P(1)*X)*exp(-(2n)^2*P(2)*X), n ->0 to 10] convol
exp[-X/P(4)]
*whew*
anyways, I've been working through the documentation on convol, and I
find it a bit cryptic. I have very few clues how to implement this
function in code. I'm guessing the first portion (the sum portion)
needs to be recursively defined in a for loop. Is this the case, or is
their a shortcut with a sigma type function built in?
However, how do I easily convolve the two functions if they are
functions and not arrays? Should I just go to fourier space?
Thanks for any help. I don't expect anyone to code this for me, just a
gentle (or violent) shove in the appropriate direction will be
infinately helpful.
|
|
|
|
| Re: I need a bit of help....Convol and functions [message #50531 is a reply to message #50455] |
Wed, 04 October 2006 06:53  |
D.Kochman@gmail.com
Messages: 7
Registered: October 2006
|
Junior Member |
|
|
kuyper@wizard.net wrote:
> D.Kochman@gmail.com wrote:
>>> Yes, func() needs to create an array for CONVOL() to convolve. However,
>>> sumex should already be an array, in order for this code to work as
>>> intended, and this code does nothing to change any aspect of sumex. I'm
>>> not sure I understand what you mean by the comment "but its a
>>> function".
>>
>>
>> Thanks for the help, slowly but surely I'm starting to get it. What I
>> meant by "but its a function" is I just don't see how sumex is an
>> array. If I were to put
>>
>> sumex = X
>>
>> that to me makes sumex a function, namely, f(x)=x
>
> No, in IDL, the statement sumex=x would make the variable named sumex
> refer to an object which is an exact copy of x. If X is an array, then
> sumex will be an array of the same type and shape. In the actual code
> you gave us,
>
> sumex=P(0)*exp( -X / P(1) )+P(2)*exp( -X / P(3) )+P(7)*exp( -X / P(8) )
>
> dividing X by P(1) creates a new array with the same shape as X.
> Passing that array to exp() produces a result with the same shape as X,
> and multiplying it by P(0) also produces an array in the same shape.
> The same applies to each of the terms in that sum, and adding all three
> terms together also produces an array of the same shape.
Aha! Therein lies my problem. You sir are a god amongst men. =)
Thanks a billion.
OK, thanks for all the help, I do understand now. WOOHOO! Sometimes
you have to realize how far you have to come in order to really
understand something. I know I don't really understand IDL, but I'm
slowly progressing. The deceptively simple syntax of IDL hides some
surprises.
|
|
|
|
| Re: I need a bit of help....Convol and functions [message #50532 is a reply to message #50455] |
Wed, 04 October 2006 06:43 |
James Kuyper
Messages: 425
Registered: March 2000
|
Senior Member |
|
|
D.Kochman@gmail.com wrote:
>> Yes, func() needs to create an array for CONVOL() to convolve. However,
>> sumex should already be an array, in order for this code to work as
>> intended, and this code does nothing to change any aspect of sumex. I'm
>> not sure I understand what you mean by the comment "but its a
>> function".
>
>
> Thanks for the help, slowly but surely I'm starting to get it. What I
> meant by "but its a function" is I just don't see how sumex is an
> array. If I were to put
>
> sumex = X
>
> that to me makes sumex a function, namely, f(x)=x
No, in IDL, the statement sumex=x would make the variable named sumex
refer to an object which is an exact copy of x. If X is an array, then
sumex will be an array of the same type and shape. In the actual code
you gave us,
sumex=P(0)*exp( -X / P(1) )+P(2)*exp( -X / P(3) )+P(7)*exp( -X / P(8) )
dividing X by P(1) creates a new array with the same shape as X.
Passing that array to exp() produces a result with the same shape as X,
and multiplying it by P(0) also produces an array in the same shape.
The same applies to each of the terms in that sum, and adding all three
terms together also produces an array of the same shape.
|
|
|
|
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
