Why does unrestricted Mo/ller–Plesset perturbation theory converge so slowly for spin-contaminated wave functions?

The poor convergence of the unrestricted Mo/ller–Plesset (UMP) series for some systems and the presence of significant spin contamination in the underlying unrestricted Hartree–Fock (UHF) wave functions have both been shown to result from the presence of low-lying, doubly excited determinants. The contributions to the UCI energy from any determinants which mix with the UHF wave function through the problematic doubly excited determinants will be described poorly by UMP theory.