I am working on the determination of protein secondary structure using FTIR. And I amfacing a challenge at the moment.
I have obtained my spectra and After performing a second derivative using Origin Pro, followed by smoothing and curve fitting, I consistently receive an error message indicating that the fit did not converge.
I would appreciate your advice on what I can do differently to address this issue and if possible some links to materials I could read r videos I can watch to help me.
Jürgen Weippert
If the fit did not converge, there are a series of possibilities:
- Your initial parameters in the fitting process are ill-chosen and don't allow the fitting process to converge towards the correct value. Origin allows you to set the initial parameters and also to define ranges. If you e.g. have a parameter that for physical reasons needs to be between 0 and 1, it makes no sense to let Origin "range free". Otherwise it might go to 20000 in the first fitting loop and will never return to a sensemaking zone.
- The fitting function is unsuitable for the data. That can also happen when there is a typo in the fitting equation, but I hope you already checked that.
- The fit went into a rotation, i.e. it ended up jumping back and forth between equally fitting parameter sets. In that case playing around with convergence criteria might be an option.
- The fitting function is overparametrized or on the verge of overparametrization for the used dataset. In the first case you are doomed, in the second case you can start by fixing some parameters to somewhat reasonable values, fitting the others, then fixing the "half-fitted" parameters and fitting the previously fixed ones. In the third step you can let all parameters roam freely and watch it converge finally.
There are more possibilities, but these are the most frequently occuring ones, I believe.
0 votes
0 thanks
- Badges
- Science topic
- Similar topics
- Mathematics
More Nicholas Alifa's questions See All
Similar questions and discussions