[gmx-users] Effect of time step size

[Mark Abraham]

On Tue, Feb 10, 2015 at 12:16 PM, Faezeh Pousaneh <fpoosaneh at gmail.com>
wrote:

> Hi,
>
> I am simulating methanol (using its topology given in Gromacs homepage, for
> 1728 molecules). I run the simulation (npt, P=1atm and T=290) for different
> time step size, dt=0.001, 0.003 fs. But I obtain different results for each
> of them:
>
>  783 kg/m^3 for dt=0.001
>  790 kg/m^3 for dt=0.003
>
> Why is it so?
>

One needs to have sampled long enough to have reached equilibrium and then
made enough observations for a sensible average. You haven't told us what
you've done there, so we'll have to assume that. The size of the time step
is closely tied to the model physics - you want to have a healthy number of
time steps per fastest motion in the system, which for atomistic MD with
some kind of bond constraints normally is at most 2fs. See manual section
on removing fast motions for some details. So if dt=3fs is reproducibly
wrong, then it looks like a case of garbage in = garbage out.

Mark


> I have also checked the point for lutidine molecule and I see similar
> behavior. Below is my production.mdb file:
>
>
>  ; RUN CONTROL PARAMETERS
>  integrator                = md
>  dt                        = 0.001       ; 2 fs
>  nsteps                    = 30000000     ; 0.002 * 50000 = 1000 ps = 1ns
>
>  ; OUTPUT CONTROL OPTIONS
>  nstxout                   = 0           ; save coordinates every 0 ps
>  nstvout                   = 0           ; save velocities every 0 ps
>  nstlog                    = 0        ; update log file every 2 ps
>  nstenergy                 = 1000        ; save energies every 2 ps
>  nstxtcout                 = 100000        ; Output frequency for xtc file
>  xtc-precision             = 100000        ; precision for xtc file
>
>  ; Neighborsearching
>  ns_type                   = grid        ; search neighboring grid cells
>  nstlist                   = 5           ; 10 fs
>  pbc                       = xyz         ; 3-D PBC
>  rlist                     = 1.0         ; short-range neighborlist cutoff
> (in nm)
>  rcoulomb                  = 1.0         ; short-range electrostatic cutoff
> (in nm)
>  rvdw                      = 1.0         ; short-range van der Waals cutoff
> (in nm)
>
>  ; Electrostatics
>  coulombtype               = PME         ; Particle Mesh Ewald for
> long-range electrostatics
>  pme_order                 = 4           ; cubic interpolation
>  fourierspacing            = 0.16        ; grid spacing for FFT
>  vdw-type                  = Cut-off
>
>  ; T Coupling
>  Tcoupl                    = v-rescale   ; modified Berendsen thermostat
>  tau_t                     = 0.1         ; time constant, in ps
>  ref_t                     = 290.        ; reference temperature, one for
> each group, in K
>  tc-grps                   = system
>
>  ; P Coupling
>  Pcoupl                    = Berendsen; Parrinello-Rahman
>  Pcoupltype                = Isotropic
>  tau_p                     = 1.0
>  compressibility           = 5e-5
>  ref_p                     = 1.0
>
>  ; Velocity generation
>  gen_vel                   = no
>
>  ; Dispersion correction
>  DispCorr                  = EnerPres   ; account for cut-off vdW scheme
>
>  ; OPTIONS FOR BONDS
>  constraints               = all-bonds   ; all bonds constrained (fixed
> length)
>  continuation              = yes         ; Restarting after NPT
>  constraint-algorithm      = lincs       ; holonomic constraints
>  lincs_iter                = 1           ; accuracy of LINCS
>  lincs_order               = 4           ; also related to accuracy
>
> Best regards
> *Faezeh Pousaneh*
> Best regards
> *Faezeh Pousaneh*
> --
> Gromacs Users mailing list
>
> * Please search the archive at
> http://www.gromacs.org/Support/Mailing_Lists/GMX-Users_List before
> posting!
>
> * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
>
> * For (un)subscribe requests visit
> https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-users or
> send a mail to gmx-users-request at gromacs.org.
>