A flat default model results in bad ringing of the reconstruction with overwhelming uncertainties in the structures. To refine the default model, the moments of the experimental data was used for the cases of the clean Ru(001) surface and the thick Pd film. For the mixed cases the reconstructions of the extreme cases are chosen for the default model.
Our experiences with applications of the maximum entropy method to ion-scattering-spectroscopy, ultraviolet-photoemission-spectroscopy and scanning-tunnelling-microscopy data show that it is far beyond a method which be used as a black box. Careful examination of the uncertainies of the reconstruction as well as the supplement of prior knowledge in the form of a soft constraint is essential for reliable conclusions. Last not least the intuition of the experimentalist sets a prior which can not be neglected.