3.62 FAQ-298 How do I set the bound for fitting for some functions?

Last Update: 2/4/2015

In the case that a fitting operation fails to converge, consider setting bounds on selected fitting parameters. Although there is no guarantee of a particular fit operation converging, the suggestions below can be useful for certain cases.

Often, once the bounded fit converges, resetting or releasing the bound(s) and fitting again will yield a converged fit as well.

The following fitting functions are mostly used in Peak Analyzer. Some of them may also be used in NLFit.

DoniachSunjic Set a small upper bound for the parameter a, i.e, a<=2, when there is too much oscillation in the fit curve following the first fitting attempt.
GaussianLorenCross Set a bound for s, such as 0.25<=s<=0.75. Reset or release the bound once the fit converges .
FraserSuzuki Fitting a negative peak is not recommended for if the x coordinate of the peak center is not positive.
Pulse This function is for fitting positive peaks. Note that the amplitude A is set to have a positive bound. For negative peak, the fitting may fail to converge.
ECS This function is more suitable for fitting positive peaks. But you can release the bound for the amplitude A for negative peak fitting.
PearsonVII This function is more suitable for fitting positive peaks. But you can release the bound on the amplitude A for negative peak fitting.
Weibull3 To get a suitable convergent fit, you may need to set w2 to a suitable value for your data, as w2 will affect the shape of the fitted curve. For example, you can fix w2=3.4 to get a shape like Gaussian. It is not recommended to free the w2. You can set a suitable bound for w2 around 3.4, for an asymmetric peak fitting.



Keywords:Peak, Fitting functions, Bound