I'm trying to fit two records with two functions that have the same parameters (3 parameters) and are symmetric to the y-axis.
It turns out that the output of the fitting over
NonlinearModelFit is very reasonable for the initial values, i. that is, when I give the baseline values that are reasonably close to my expectations (by analyzing the functions and the system), the fitting works and I get reasonable results. But if I drastically change the starting values, the results are completely unacceptable.
Am I doing something wrong or is it more likely to happen? In other words, this adaptation process is w.r.t. unstable? to select the starting values, d. H. you can not choose the initial data arbitrarily?
What is the scope of the initial values?