[gmx-users] shuffle option

Lubos Vrbka lubos.vrbka at gmail.com
Thu Jan 19 15:45:49 CET 2006


> just the order of the molecules is mixed up. as long as you stick with 
> this ordering it is fine. but if you want to go from e.g. 12 to 6 
> processors you may need to reshuffle.
ok. the relative ordering is not very important for me, but it could...

this all means that if i perform the same simulation several times (e.g. 
i have 1ns chunks that i repeat till i am satisfied with the results), i 
should shuffle only in the very first run.
if i do it in every run, what happens if i load several such 
trajectories to,e.g., vmd program? on the same number of processors it 
might remain the same, however if the number of processors change, will 
then atom/residue numbers be in disagreement in different chunks of the 
whole trajectory?

> anyway, shuffle can be undone but not sort.
so, in my case, i shouldn't use sort at all to keep the ordering the same?

> in gromacs 4.0 these options will disappear and we'll have domain 
> decomposition.
looking forward to it :)

thanks,

-- 
Lubos
_ at _"



More information about the gromacs.org_gmx-users mailing list