Permalink of this document: DOI 10.3247/SL5Nmr14.002.
Please, cite this online document as:
Sykora S., Fitting of spin system parameters to NMR spectra revisited,
29th Valtice NMR, Valtice (Czech Republic), 27-30 Apr 2014, DOI: 10.3247/SL5Nmr14.002.
Simulation of NMR spectra, and fitting of spectral parameters to actual experimental data, are time-honored tasks in NMR spectroscopy. The basic principles are well known and, in principle, extensively tested.
Historically, however, several practical obstacles were encountered and only partially overcome, the most important of which being the extreme computational power required even for quite small spin systems and, in fitting, the presence of many multiple local minima and false solutions. These obstacles, alleviated a bit - but by no means overcome - by modern computer technology, become particularly critical in spectra with dynamic effects (such as chemical exchange).
With increasing size of molecules of interest, the numbers of transitions to be handled become soon astronomical and this raises the nearly unsurmountable problem of assigning transitions to spectral peaks (which are relatively few). Even though techniques such as integral transforms have alleviated these concerns, even those are still burdened by the need for very clean and pre-edited spectra.
The mounting need for fully automated spectral analysis of realistic spectra (i.e., ones burdened by the presence of solvent and impurity peaks, and artifacts) of molecules with up to 100 spins raises new problems, both in the area of simulation methods and their execution times, and in the area of automated pre-editing of the experimental data.
This presentation will give an overview of several approaches to tackling these problems.
Your comments are welcome and will appear here