Re: Doubt in polynomial fitting - emergency [message #73345] |
Tue, 02 November 2010 03:38  |
sid
Messages: 50 Registered: January 1995
|
Member |
|
|
On Nov 2, 2:22 pm, Bringfried Stecklum <steck...@tls-tautenburg.de>
wrote:
> sid wrote:
>> Hi,
>> I am fitting my spectral data with 2 degree polynomial with the
>> routine svdfit. I need to find the minmum value after fitting the data
>> points.
>> For example my code is like this,
>> ;c is x axis with wavelength(it is an absorption line)
>> ;d is y axis with normalised intensity
>> ypoly=svdfit(c,d,3,yfit=y1,chisq=chi,sigma=sig)
>> x=min(y1)
>> x1=-(ypoly(1))/(2*ypoly(2))
>> if suppose u,v is the position of x, x1 respectively then
>> c(u) should be equal to c(v)
>> and
>> d(u) should be equal to d(v)
>> but im getting
>> c(u)=3933.3090 in angstroms
>> c(v)=3933.3072 in angstroms
>> d(u)=0.071168385
>> d(v)=0.072779992
>> Please suggest me why it is not the same, which value should I believe
>> and how?
>> thanking you
>> sid
>
> Well, I'd say x represents the minimum for the fitted data while x1 is the
> minimum for the fitted polynomial. Since the spectrum is sampled at discrete
> points c, you cannot expect that the abscissa value for the minimum of the
> polynomial coincides with a sampling point.
>
> Regards, Bringfried
Which value will be the best and how to find that, by looking into the
plot I feel x is good.
thanking you
sid
|
|
|