[gmx-users] shells and VdW interactions

[Stefano Piana]

I am currently running shell MD in an ionic crystal and with version 3.2 
it runs fine, however for my system I have to use a rather small 
timestep to keep energy conservation.
I would like to ask if anybody knows if it is possible to put a Buck 
potential on the shells and how is this accounted for in terms of 
exclusions.
Can I define pairs between shells and shells and/or between shells and 
atoms?
With version 3.2 the buckingham potential seem tu work fine, however
I observe a weird behaviour when I try to build pairs with a Buckingham 
potential; although my C term of the Buckingham is 0 the total 1-4 
interaction energy is large and negative (!!). Furthermore I observe 
that after a few steps the 1-4 interacting atoms collapse and the system 
blows up.
I did not have time to check throughly all the numbers but it looks to 
me as if there is some kind of sign problem in the creation of the 
pairs; as a crosscheck I changed the sign of the 1-4 scaling factor to 
negative and now the simulations runs fine (Although I would not trust 
the results at all...).
Suggestions? Comments?
Bye,

Stefano