Position and momentum densities complementarity at work, Refining a quantum model from different data sets
J.-M. Gillet,a S.
Ragot and P.J. Beckera
aLaboratoire
Structures, PropriŽtiŽs et ModŽlisation des Solides, CNRS/UMR8580, Ecole
Centrale Paris, Grande Voie des Vignes 92295 Chatenay-Malabry Cedex, France (gillet@spms.ecp.fr)
Although
Bragg and Compton scattering are well-established techniques, very few attempts
to combine information originating from those experiments have been made so
far. This remark also holds for Bragg magnetic scattering.
We
propose a quite general procedure to refine a quantum model from different data
sets using basic Bayesian statistics. As an illustration, different results for
extracting chemical information such as charge transfer in ionic-covalent
compounds will be given.
Influence of correlation in data, as introduced in momentum density reconstruction, will be explored and extension to other combinations of experiments will be suggested.
At a
different level of interpretation and accuracy, a simultaneous approach of
position and momentum representations of the One electron Density Matrix (1DM)
requires, at least, a possibility of working beyond the straight path of
Òindependent electronsÓ scheme in a perfect 3D lattice. We will describe an
ab-initio computational strategy to the calculation of the 1DM that reproduces
infinite lattices results with a good accuracy but that enables for the of
defects in the structure as well as a fair account for electron correlation
energy. This Cluster Partitioning Method (CPM) will be shown to be efficient
for studies ranging from intra-atomic correlation mechanisms in ionic solid to
analysis of hydrogen bond in ice and, with few compromises, an extension to
more complicated structures.