Hello, I already posted in the forum with some problem about fitting Pade Aproximants.

Now I see where the problem it is, I will explain to see if somebody may help:

I was fitting using some Pade Aproximants expansion.

In each new Pade Approximant, I took the previous parameters as starting value, and did a scan in a for{}; loop to get a good value for the new parameter.

They were working fine for the first 3 approximants, however, in the 4th one, Minuit found a minimum which was not the best one, and for the 5th one it’s even worse.

I found the origin of the problem by using Mathematica:

I gave Minuit the solution from Mathematica and it worked ok, however you need to give Minuit the parameters with high precision.

I realized by plotting the function that this was due to the fact that the function is quite sensitive to the parameters (i.e. if some parameter is “-0.205074”, and you round it to “-0.20507”) the function behaves quite different, so you need great precission.

Do you wonder how could I solve this problem or should I give up with Minuit?

I attach the program “Myfit.C” and the data “logdata.C” which should be renamed to “logdata.dat”