CCL: QM/MM cutoffs

Hi folks,
 I'm currently implementing a simple QM/MM molecular dynamics approach,
 but I have some doubts about the right way of dealing with the
 electrostatic interactions between the quantum distribution and the
 classical point charges. Using periodic boundary conditions and the
 minimum image convention, I have to define a cutoff distance : a
 site-site interaction is discarded if the site-site distance is greater
 than this value. OK for a classical simulation.
 Now if part of the system is treated quantum mechanically, the
 definition of some QM sites is non-obvious : the QM wavefunction is
 delocalized on the whole QM subsystem.
 The simplest idea would be to select the center of mass of the QM
 subsystem as a unique site only to choose MM point charges interacting
 with the QM subsystem, but if the QM molecular shape is not spherical,
 this solution is not convenient (a QM atom located at one end of the QM
 subsystem could not interact with the closest images of some MM point
 charges).
 Thus I'd like to hear about your solutions/experiences/references to
 solve this problem, taking in mind that I don't want to approximate (for
 the moment) the QM electronic distribution (no fitted multipoles ...) I
 know about Ewald sums for QM/MM interactions but I'd prefer to start
 with a rather simplest approach.
 Best regards,
 Nicolas
 --
 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
 Nicolas FERRE' (PhD)
 				 phone/fax : +39-0577-234278
 Dipartimento di Chimica
 Universita` di Siena             mailto:ferre^at^unisi.it
 via Aldo Moro
 53100 SIENA (Italia)             http://ccmaol1.chim.unisi.it/
 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~