From: Jérôme Hénin (jhenin_at_ifr88.cnrs-mrs.fr)
Date: Mon Mar 28 2011 - 15:50:44 CDT
On 28 March 2011 18:36, Bjoern Olausson <namdlist_at_googlemail.com> wrote:
> For now my bins differ from 60 to 70000 Structures after 10ns.
>
> I'll see what happens after 100ns. But since I doubt there is a better colvars
> selecction I guess I have to deal with slow-relaxing, orthogonal degrees of
> freedom.
>
> Or would you suggest another colvar selection?
> I'll shortly summarize my setup:
> Homodimer with 2423 Atoms the linker between the monomers is only a few ~5
> Aminoacids long. For the sake of speed I have only choosen the C-Alphas and
> the according Protons since I am using shake so my setups is as follows:
You can save atoms (hence time) by using carbonyl carbons instead of
the CA-HA pair. Actually, recent tests indicate that you could just
remove the HAs from your current setup and things would be fine, i.e.
the "orthogonality to constraints" requirement apparently does not
apply to the current code (unlike early implementations, constraint
forces seem to be accurately accounted for - however the theory for
this has not been properly derived yet). I recommend the carbonyl
trick.
And since you are dealing with large motion, slow diffusion could be a
large part of the problem. Splitting things into a few windows is
likely to help at least a bit.
Cheers,
Jerome
This archive was generated by hypermail 2.1.6 : Mon Dec 31 2012 - 23:20:01 CST