An illustrative piece of literature would be: http://pubs.acs.org/doi/ In this paper, magnetic parameters are derived using the 'Broken Symmetry Approach' popularised by Noodleman. In this work, the true singlet, which would be found by a CI of the two broken symmetry wavefunctions – Cu(1/2)---Cu(-1/2) with Cu(-1/2)---Cu(1/2) - is approximated from just one of these functions. The CI is, of course, too difficult to perform. To tackle this problem, we would use Metadynamics (http://journals.aps.org/prl/) to find both broken symmetry states, tracking them across the geometries of interest. We would then be in a position to perform a NOCI calculation (http://scitation.aip.org/) to yield a better guess at the true singlet. It would, for example, not experience spin contamination. An interesting extension is then to use LOBA (http://pubs.rsc.org/en/) to help understand how these magnetic orbitals are formed from the atomic basis.