EVB/MM

In the Empirical Valence Bond method (EVB) [112,113] from Warshel and co-workers a chemical reaction is described using a valence bond approach. The system wavefunction is represented by a linear combination of the most important ionic and covalent resonance forms and the potential energy is found by solving the related secular equation. The electronic interaction Hamiltonian is built using parameter terms extracted from empirical values and/or ab initio surfaces. So, this fact must be emphasized, EVB approach is built as a sum of parametric functions very similar to a MM force field without any explicit treatment of the electrons. This makes the method usually tens of times faster than the QM(SCF)/MM method.

The main advantage of the EVB approach is its ability to give at low computational cost very good quantitative results in comparison with experiment as long as the incorporated empirical terms are carefully chosen[114,115]. This is mainly accomplished by first calibrating free energy surfaces from reference reactions in solution before incorporating the enzyme effects. However, the main disadvantage is that choosing the valence bond forms (i.e., the most prevalent ionic and covalent forms), one implicitly directs how the chemical reaction should be held. Therefore the EVB method cannot allow unusual reaction pathways that could occur in reactive chemical systems or a chemical reaction not previously defined in the valence bond forms.

Other Alternatives:
Hybrid methods are continually in expansion. Some methods are based on valence bond approximations such as the MOVB from Mo and Gao[116], where all configurations are calculated ab initio. On the same valence bond framework SEVB approximation is designed to correct a PES previously calculated [117]. Another option is the Warshel's frozen or constrained DFT (CDFT) [118] which basic idea is to treat the entire protein solvent system quantum mechanically while freezing (or constraining) the electron density of the environment

The QM/MM framework has had many different implementations. An example of them are the overlapping mechanically embedded [119], the effective charge operators [120], the effective group potential[121] or the effective fragment potential[122]. Other interesting but yet not fruitful strategies are devoted to couple two wavefunctions at two different QM levels polarizing each other in a self consistent fashion [123,124,125,126].

In my opinion these wide variety of QM/MM methods that address to the same problems must converge to a unique and standardized procedure. Hopefully, linear scaling techniques will permit one day to study big systems without artificial partitions and ambiguous embeddings.