### Dmfit Program - Home Page

The Dmfit programs enables fitting of solid state (and liquid) NMR spectra, including 1D and 2D datasets current version is : dmfit#20190110 zip [Download...
] |

The "dmfit" program is is a versatile tool designed for fitting NMR spectra, including solid state NMR. It provides a variety of different models that account for :

- Gaussian/Lorentzian lines,
- spinning sidebands,
- Chemical Shift Anisotropy (static and MAS),
- first and second order Quadrupolar interaction (static and MAS)
- ...

**New** : 64 bits version for Windows, report bugs on the Bug report page...

#### Optimisation Algorithm

The mismatch criterion that we use is a simple quadratic distance between the experimental and modelled spectrum. This could of course be improved in some cases but its advantage is to make no hypothesis upon the type of data that we consider. The optimisation algorithm aims at minimizing the mismatch criterion by varying the parameters of the different lines. As the computation of the model includes non-linear dependencies to the different parameters and, in some cases, no simple analytical form, we use an iterative algorithm. Starting with an approximate solution we compute and add an increment to try to converge to a local minimum. This does not ensure the unicity of the obtained solution that will only be the closest to the starting point. The algorithm used in our program is a constraint gradient protocol which involves computation of the partial derivatives over each of the parameters for all variable parameters and each point of the experimental spectrum. The resolution of the n-linear system of equation yields an increment that is finally added to the starting parameter set. This procedure is rather robust when close to a final well defined.

#### Acknowledgements

**Massiot et al.**Magnetic Resonance in Chemistry, 40 pp70-76 (2002) [see more or ask for a reprint]

ISI citation index 2363 (09/01/2019) - 7551 users with download

**"One of the most cited recent paper in the field of Engineering" ISI July 2003**

Communiqué de Presse CNRS

*we thank all "dmfit" users for this achievement*