Full vs. constrain geometry optimization in the open-closed method in estimating the energy of intramolecular charge-inverted hydrogen bonds

It seems that the most reasonable reference open system, in the closed-open method of estimating the energy of the intramolecular hydrogen bond, should have geometry parameters optimized but with the simultaneous freezing of those geometry parameters whose relaxation may lead to the significant change in the bonding pattern. Thus we propose that the reference system should be obtained after the constrain geometry optimization rather than after the geometry full optimization.