Cooperative versus dispersion effects: what is more important in an associated liquid such as water?

We implemented the quantum cluster equilibrium theory in our postprocessing program PEACEMAKER. This program may be run in conjunction with the very efficient vibrational frequency analysis code SNF and can therefore provide access to all electronic structure programs combined with this program. We applied the quantum cluster equilibrium theory in order to investigate the influence of a wide range of electronic structure models on the description of the liquid state. This investigation revealed much about the relevance of approximations in modern simulations of associated liquids such as water. While it is often claimed that the use of density-functional theory in condensed matter is leading to gravely erroneous results, we found that, contrary to these assertions, the exact exchange functional B3LYP and the gradient-corrected functional BP perform very well in combination with sizable basis sets as compared to second-order Moller-Plesset perturbation theory employing the same basis set. The use of density-functional theory with smaller basis sets does, in fact, lead to better results in the liquid state than the use of second-order Moller-Plesset perturbation theory in combination with these small basis sets. Most importantly, the neglect of cooperative effects disturbs a good description much more evenly if we apply second-order Moller-Plesset perturbation theory in combination with large basis sets than density-functional theory including cooperativity with smaller basis sets or Hartree-Fock using a very small basis set.