[gmx-developers] ensemble code and restraints

Berk Hess hessb at mpip-mainz.mpg.de
Tue Dec 9 10:42:00 CET 2008


I don't understand why you want to ensemble restrain only over pairs
(and I don't have time to read papers).
Ensemble restraining mimics real NMR experiments where
you have a large number of molecules in a test tube and
you measure an ensemble averaged signal.
Such signals are r^-6 averaged over all pairs.
Because of the extremely non-linear averaging, if one pair
out of 10 is a bit below the upper bound, there will be no
force on all 10 pairs.


Robert Fenwick wrote:
> OK I am still a bit lost about the restraining of distance restraints
> over pairs of structures within an ensemble. I have looked at the
> manual and it seems that what I would like to do is indeed ensemble
> averaging over pairs. How is it possible to specify which replicas are
> involved? I essentially want to treat these pairs as dimers, so that
> there is only a force on the molecules if both structures in a pair do
> not fulfill a distance restraint.
> The system suggested by Richter et al. is a chain loop, i.e. pairs of;
> 1 and 2, 2 and 3, 3 and 4, 4 and 1
> Cheers,
> Bryn

More information about the gromacs.org_gmx-developers mailing list