Hello,
I wonder, why in the QM/MM model it is advisable to neglect bonded interactions at the QM-MM boundary that involve just one(!) MM atom (i.e. all bonds of the type QM1-MM1, bends of the type QM2-QM1-MM1, and torsions of the type QM3-QM2-QM1-MM1)? For instance, this is set up by default in ORCA. But what is the reason behind this? I would appreciate very much references to papers where this issue has been addressed.
Best regards,
Evgeniy