Large error on parameter

Dear experts,

I am fitting a data with two types of functions:

f1 = sqrt( p1^2/x^2 + p2^2 ) f2 = sqrt( p1^2/x^2 + p2^2/x + p3 )

Please find the plots attached. fit_par2.png for f1 and
fit_par3.png for f2.

As can be seen f1 does not fit good at high values but f2 does. I am interested in the parameter without ‘x’.
With f1, the parameter with ‘p2’ has small error.
The problem with f2 is that the error on parameter ‘p3’ is quite large. And this is (p3) the parameter I am interested in.
Is there a way to reduce the error on this parameter? I tried changing the initial values but it seems of no help - it still gave large error on the last parameter.


You can try to play with some fit options (e.g. try “E”, “ME”, “L”) but it seems to me that your current fit suggests that the last parameter in “f2” is compatible with 0, so you can try:
f3 = sqrt( p0/x^2 + p1/x ) = sqrt( (p0/x + p1)/x )

BTW. Maybe you should fit the “integral in bin”, so try to play with “I”, “IE”, “IME”, “IL” fit options, using your “f1” and “f2” functions.