| help regarding curvefit [message #89833] |
Mon, 08 December 2014 22:56  |
Krishnakumar M.A
Messages: 19
Registered: March 2013
|
Junior Member |
|
|
Hello All,
I am trying to use the curvefit algorithm in IDL for fitting my data with a function as given in the code below. It is peacefully fitting and giving me the result, but only one problem, it is doing only one iteration and giving the same A=[] value as the fitted result. I'm not able to figure out what went wrong, and I have used cuurvefit before and it worked well. Do anybody have any idea to solve this issue?
Any help is appreciated....
Thanks,
Krishnakumar
code:
RO gfunct, X, A, F, pder
F = sqrt((!dpi^5.0 * A[0]^3.0) / (8 * x^5.0)) * exp( - (!dpi^2 * A[0]) / (4*x)) * 180.0
IF N_PARAMS() GE 1 THEN $ ; calculate the partial derivatives......
pder = [!dpi^5.0 * A[0] / (8*x^5.0) * exp(-!dpi^2.0 * A[0] / (4*x)) * 180.0* (1.5 - !dpi^2.0 * A[0]/ (4*x))]
END
openr,1,"data.dat"
xpy=fltarr(3,301)
readf,1,xpy
x=reform(xpy[1,*])
y=reform(xpy[2,*])
A=[100.0]
weights = 1.0/y^2.0
yfit=curvefit(x+10,y,weights,A,SIGMA,chisq=chisq,iter=it,fun ction_name='gfunct')
print,A,sigma,chisq,it
close,1
plot,x,yfit
oplot,x,y
end
|
|
|
|
|
|
|
|
| Re: help regarding curvefit [message #89840 is a reply to message #89835] |
Tue, 09 December 2014 23:34  |
Krishnakumar M.A
Messages: 19
Registered: March 2013
|
Junior Member |
|
|
On Tuesday, December 9, 2014 2:24:52 PM UTC+5:30, Heinz Stege wrote:
> Hello Krishnakumar,
>
> please doublecheck the partial derivative.
>
> Your function is
> F(A) = const * G(A)
> with
> const = 180 * SQRT(!dpi^5/(8*x^5))
> So your partial derivative should be
> dF(A)/dA = const * dG(A)/dA
> The SQRT seems to be missing in your derivative.
>
> HTH, Heinz
Thanks Heinz. That was a silly mistake I did not notice.
Now the code is working.
Krishnakumar
|
|
|
|
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
