Basically you should know the answer before you start fitting. We recommend doing local fits at several wavelengths to get a general idea of the kinetic model you have. Then number of the components in the Global Fit will correspond to the number of kinetic components (lifetimes) you expect to see in all of your data. Surface Xplorer will do its best to find their amplitudes, lifetimes and confirm your suggestions by displaying the approximation errors.
At the SVD step (Principal components via SVD) you need to check this number using singular values that determine the significance of each component. The principal components are sorted in decreasing order. Of course, the more components you choose, the more details (and more noise) you take into account. Stop when singular value of the last component is much smaller than that of the first one.
At the Global Fit step you will decompose the principal components into exponentials. The number of components (with different lifetimes including an infinite) in the fit usually matches the number of significant principal components.
Generally the above three numbers should match. If they don’t then possibly:
The time zero and the IRF are different for different wavelengths. Optimizing the chirp may help in this case. Please check kinetic fits at different wavelengths. It may help to fix these parameters when performing the Global Fit.
The lifetime of a component changes with wavelength. It may help to fix this lifetime when performing a Global Fit.
A component is present only in a short range of wavelengths or it is very fast and its influence is small. You may need to limit the range where you perform the SVD decomposition and the Global fit.
Too many components are set, it does not always increase the quality of the fit.
This site is protected by reCAPTCHA and the Google Privacy Policy and Terms of Service apply.
© 2002 – 2024 Ultrafast Systems, 8330 Consumer Ct, Sarasota, FL 34240, USA. All Rights Reserved.
An answer to your question might already be there.
