[gmx-users] Should I use continuation = yes in my production dynamics following equilibration?

[Justin A. Lemkul]

On 5/13/12 9:11 PM, Andrew DeYoung wrote:
> Greetings,
>
> I am running 5 ns of equilibration followed by 10 ns of production dynamics.
> Equilibration and production dynamics differ only in the number of steps and
> the frequency of saving the positions, velocities, forces, and energies.  To
> conserve disk space, I set nstxout = nstvout = nstfout = nstenergy to some
> high value, such as 1000000 (in equilibration.mdp), for equilibration.  Then

You can set them to zero to completely eliminate output, if you need to.

> for production dynamics, I set nstxout = nstvout = nstfout = nstenergy to
> some relatively low value, such as 100 (in dynamics.mdp), because it is from
> these data that I will do property analysis.
>
> So, to do this, I have been using a workflow like this:
> (1) For equilibration, feed equilibration.mdp to grompp using "-f".  Then
> run mdrun, saving the checkpoints to state.cpt.
> (2) For production dynamics, feed dynamics.mdp to grompp using "-f".  Also,
> feed state.cpt to grompp using "-t".
>
> where equilibration.mdp and dynamics.mdp contain different values of nsteps
> and of nstxout, nstvout, nstfout, and nstenergy.
>
> My question is, to do this sort of continuation (where I'm ONLY changing
> nsteps, nstxout, nstvout, nstfout, and nstenergy), should I use
> "continuation = yes"?
>
> The manual gives this description of continuation:
>
> "This option was formerly known as unconstrained_start.
>
> no: apply constraints to the start configuration and reset shells
>
> yes: do not apply constraints to the start configuration and do not reset
> shells, useful for exact coninuation and reruns"
>
> I do not think that my system has any constraints, since all the bonds in
> all molecules are flexible, with force constants specified in the .itp
> files.  If this is the case, then should I use continuation = yes?  Or,
> since I do not have constraints in my system, will it perhaps not make a
> difference whether continuation is yes or no?  Do you have any
> recommendations?
>

If not using constraints, the continuation keyword is irrelevant.  The key is to 
be sure there are truly no constraints, including any that use SHAKE or LINCS 
(i.e. a solute) or SETTLE (water).  The water models present in Gromacs should 
all be rigid (i.e. constrained using SETTLE).  But if you're sure there are no 
constraints and that this representation is proper, then you can ignore the 
"continuation" keyword.  Saying you "think" there are no constraints means there 
is uncertainty, which means you have some checking to do ;)

-Justin

-- 
========================================

Justin A. Lemkul, Ph.D.
Department of Biochemistry
Virginia Tech
Blacksburg, VA
jalemkul[at]vt.edu | (540) 231-9080
http://www.bevanlab.biochem.vt.edu/Pages/Personal/justin

========================================