[gmx-users] questions related to md.mdp

[Peter Kroon]

Hi,

On 06/04/16 10:22, Brett wrote:
> Dear All,
>
> Based on the on-line lysozyme tutorial,
>
> the first several lines I used for my md.mdp was as following,
>
> title        = OPLS Lysozyme MD 
> ; Run parameters
> integrator    = md        ; leap-frog integrator
> nsteps        = 5000000    ; 2 * 5000000 = 10000 ps, 10 ns
> dt        = 0.002        ; 2 fs
> ; Output control
> nstxout        = 1000        ; save coordinates every 2 ps
> nstvout        = 1000        ; save velocities every 2 ps
> nstenergy    = 1000        ; save energies every 2 ps
> nstlog        = 1000        ; update log file every 2 ps
>
> My first question is, besides 2 fs for the dt, can I change dt to 20 fs or 200 fs, or 2000 fs for very long production MD? And what will be the difference for the results of production MD between dt=2 fs and dt=2000 fs?
Not with impudence. The larger the timestep, the larger the error in the
numerical integration (from acceleration to displacement). In practice
this means your results will be unphysical, and your simulation will
crash/explode. Essentially you make the assumption that the force
remains constant over dt when doing the integration.
>
> My second question is, the intervals for saving coordinates, velocities, energies, log file should be same, or can the intervals different?
They can differ.
>
> My third question is,  the intervals for saving coordinates, velocities, energies, log file should be an integer times the dt, or have no relationship with dt?
This should be an integer, and it represents when (in number of steps)
the quantity is saved. This means that if nstlog=1000 and dt=2fs, you
write something to the log file every 2 ps, because 100*2fs = 2ps.
>
> Brett
Peter