[gmx-users] RE: logfile size

[Mark Abraham]

On 7/12/2011 7:41 PM, lloyd riggs wrote:
> Message: 3
> Date: Tue, 06 Dec 2011 18:58:50 +0100
> From: Javier Cerezo<jcb1 at um.es>
> Subject: Re: [gmx-users] RE: logfile size
> To: Discussion list for GROMACS users<gmx-users at gromacs.org>
> Message-ID:<4EDE57DA.80405 at um.es>
> Content-Type: text/plain; charset=UTF-8; format=flowed
>
> Not the main problem, but your dt=0.0002 (0.2fs), is it what you want?
>
> Javier
>
>
> Yeah,
>
> I found when you have 10,000 + protein atoms + assorted others (2 different ligands of say 5000K atoms) the time steps have to remain smaller or you start getting problems with all sorts of things from Temp coupling to simple integrations between steps.  I can EQ it at 0.02, for 1M steps or more, and still run into the same problem when I switch to MD integrators.  Dont know if this is universal though, as I spent time making sure the models are correct, ie error free after EQ's (em, NVT, NPT).  They run though fine at that time step, it just makes the whole simulation take a while to run, as time steps translate into direct computational time.
>

Post-equilibration MD at normal T and P with all bonds constrained 
should be stable with 2fs time steps. If not, then I'd say there's 
something wrong with the parameters and a smaller time step is not 
actually a fix.

Mark