Talk:QM/MM

suggested changes

This article is a stub right now. —Preceding unsigned comment added by Chibibrain (talk • contribs) 23:45, 5 May 2009 (UTC)[reply]

Some established popular concepts/methods can be mentioned:

1. polarizable/electrostatic/mechanical embedding.

2. different implementations of the QM/MM term in the Hamiltonian, such as the electrostatic term, the van der waals energy term etc.

3. the methods for dividing a molecule into MM and QM regions, such as link atoms, hybrid orbitals, etc.

The popularity of QM/MM is on the rise, as gleaned from the new developments in force field simulation methods (AMBER, CHARMM) that incorporate quantum chemistry methods (semi-empirical and first principles). There are also recently developed stand-alone supervisor programs that runs and communicate a QM package and a MM package, such as PUPIL by Prof. Roitberg's lab at U. Florida, and QMMM by Prof. Truhlar's lab at U. Minnesota.

Chibibrain (talk) 00:03, 6 May 2009 (UTC)[reply]

Prof. Levine's thesis on QM/MM is also freely available online from his website. A link could be provided.

Chibibrain (talk) 00:09, 6 May 2009 (UTC)[reply]

Some criticism to a citation from the article that doesn't concern the QM/MM methods themselves: "On the other hand the simplest ab-initio calculations typically scale O(N3) or worse (Restricted Hartree–Fock calculations have been suggested to scale ~O(N2.7)"

This is not generally true. For systems described by a sparse one electron-density matrix (and fock/kohn-sham matrix, resp.), the cubically scaling diagonalisation step in Hartree-Fock theory and Kohn-Sham density functional theory can be avoided, and also the fock matrix can be built with only linear scaling effort. This especially applies to the biomolecular systems (proteins, DNA, etc.) typically studied by QM/MM methods. Thus, what makes QM/MM methods cheaper than pure ab-initio methods for a given system are the the prefactors in the effort. These also form the reason why the (cubically scaling) diagonalisation procedure is often cheaper than O(N) methods (up to a system size of rougly 5.000 - 10.000 basis functions). —Preceding unsigned comment added by 134.2.64.110 (talk) 14:38, 15 December 2010 (UTC)[reply]

commercial?

What kind of commercial is this?? I am not an expert but it seems these are chosen quite randomly (added themselves perhaps?). [quote] "...it has been adapted by several groups including (but not limited to): Weitao Yang (Duke University), Sharon Hammes-Schiffer (The Pennsylvania State University), and Kenneth Merz (University of Florida)." [/quote] — Preceding unsigned comment added by Jaapkroe (talk • contribs) 14:17, 31 January 2011 (UTC)[reply]