[gmx-users] cvs version and position restraints...

[Andrea Vaiana]

Hi everyone,

I have a small DNA duplex in a dodecahedron box. Amber forcefield. Spce 
water. I have position restraints defined for the whole duplex in my 
topology. I'm running exactly the same input script on 2, 4 and on 12 
cpus. The one with 12 produces a trajectory in which it is quite clear 
that the restraints are working: the stronger the restraints on a given 
set of atoms, the less these atoms move   away from the equilibrium 
position during the simulation. With 4 and 2 cpus, it is af if no 
restraints are used (the whole system tumbles and moves around the box), 
although I do see energy terms for the restraints in the log file...
I'm using the CVS version (downloaded June 4th).
I'm not sure but I think I remember someone having a similar problem on 
version 3.2 but I don't remember how/if it was solved. I couldn't find 
anything on the mailing lists.
Any clue about what's going on?

Thanks,
Andrea