LargeSpectra

Handling of Large Spectra

The INFOS fitting algorithm is able to handle fitting of large spectra with many peaks by breaking the spectrum into sections and fitting each of these. Breaking the spectrum into small sub-spectra makes the individual fit optimization simpler, and therefore less demanding on memory and computer time. However, if sections become too small, peaks overlap between sections and fitting becomes worse, since fits of individual sections become highly interdependent. The FitSpec function calculates sizes of the sub-spectrum, which is optimized for most spectra. However, spectra with many peaks by default have smaller sub-spectra, which makes fitting of broad peaks more difficult. In the case that a large range of linewidths exists in a spectrum, then may be useful to decrease the number of sub-spectra from the default, using ‘par.grid’.