Hello Rooters,
as I know the chi-square method, used to estimate parameters, goes like
this:
chisquare=sum_{i=1}^N (x_i - y_i)^2/e_i
with x_i being the value from a spectrum and y_i being the theoretical
function depending on one or more parameters.
To get the best-fitting values for the parameters this function must be
minimized.
Suppose to fit a constant y with fit-option 'w' (setting all errors to 1).
Then the chisquare-value should be calculated like this:
chisquare=sum_{i=1}^N (x_i - y)^2
It is simple to calculate y for which this value is minimal:
y=( sum_{i=1}^n x_i ) /n
so y is simple the mean value of the data in the spectrum.
I checked this with the following program:
TH1F* h1=new TH1F("h1","test",10,1,10);
h1->SetBinContent(3,1);
h1->SetBinContent(7,2);
h1->Fit("pol0","W");
but what I get is a best-value of 1.5 instead of 0.3!
It seems that ROOT ignores every channel with zero-content.
Is ROOT ignoring channels with zero-contente and why?
What can I do that ROOT gives the correct value?
Cheers
Alexander Dietz
This archive was generated by hypermail 2b29 : Sat Jan 04 2003 - 23:51:18 MET