[gmx-users] Effect of time step size

Faezeh Pousaneh fpoosaneh at gmail.com
Wed Feb 11 10:37:34 CET 2015


Dear Justin,

Thank you.
You are right about we must check validity of typologies attached in
Gromacs webpage, but I just took it for test of influence of 'time step' on
results when I was faced with similar problem in my own defined molecule. I
initially was simulating my Lutidine molecule, so let me please give
details of it, (running time 15 ns with same .mdp file as I sent before
except this time both with Parrinello-Rahman, 230 molecules in rectangular
box):

dt=0.002

Energy                      Average   Err.Est.       RMSD  Tot-Drift
-------------------------------------------------------------------------------
Temperature                 289.953       0.03    5.63283   0.189132  (K)
Pressure                    3.70721        2.6    389.728  -0.136095  (bar)
Density                     956.957       0.67    8.57377    3.32218
 (kg/m^3)

dt=0.001

Energy                      Average   Err.Est.       RMSD  Tot-Drift
-------------------------------------------------------------------------------
Temperature                 289.994      0.033    5.70319  0.0487858  (K)
Pressure                    1.82689        1.2    376.771    2.81752  (bar)
Density                     967.363       0.58    8.21492    2.86186
 (kg/m^3)


Yes, I let time for relaxation and I ignore those times before
equilibration. I do determine time-dependence of parameters like density
and etc, but also I check RMSD of lutidine-lutidine in order to carefully
collect the data which are stable. But as you see still there are
difference in densities? Isn't is big difference?

Many thanks
Best regards
*Faezeh Pousaneh*

On Tue, Feb 10, 2015 at 9:49 PM, Justin Lemkul <jalemkul at vt.edu> wrote:

>
>
> On 2/10/15 3:41 PM, Faezeh Pousaneh wrote:
>
>> Dear Mark,
>>
>> Thank you so much for the reply.
>> I had sent the .mdp file in my previous email (below). In more details I
>> had equilibriuted 1728 methanol (NPT ensemble at T=290 and P=1atm, using
>> all-bond constraint) for long enough time (10 ns) and I am sure I have
>> reached the equilibrium according to RMSD analysis.
>>
>>
> RMSD of what?  I doubt that analysis of any value here.  You should
> determine the time-dependence of quantities of interest, like diffusion
> constant, density, etc.
>
>  1- You are right about time step with constraint, however I have been
>> running for dt=0.002 as well, and I get density 790 kg/m^3. it means that
>> still there is a big difference in result of dt=0.001 and dt=0.002:
>>
>>>   783 kg/m^3 for dt=0.001
>>>   790 kg/m^3 for dt=0.002
>>>
>> isn't big difference? then how it is said that dt can be at most 0.002?
>>
>>
> These are single numbers without error estimates, so there's nothing to be
> said.  If both values are +/- 20, then there's probably no difference.  Are
> you allowing time for relaxation, or are you including all frames (thus
> some unequilibrated time)?
>
>  2- I see that in the experiment the density of ethanol is 791 Kg/m^3 at
>> T=290. That is not in agreement with my result for dt=0.001. What does it
>> mean? does it mean that the given topology of ethanol (I take it from
>> Gromacs homepage) is not well defined?
>>
>>
> What topology is that?  I see methanol for download, but not ethanol.
> What is the reference for the parameters?  If there isn't one, I'd be
> skeptical about using it.  Find a model that has a known (expected) density
> value, and make sure you can reproduce it using sensible run settings.
> Note too that "from the GROMACS homepage" doesn't necessarily imply
> validity; the user contributions are provided as a convenience but there is
> no endorsement of correctness; that's for you to evaluate based on
> publications and your own assessments.
>
>
>    integrator                = md
>>   dt                            = 0.001
>>   nsteps                    = 10000000
>>   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             ;
>>   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
>>
>
> The Berendsen algorithm does not yield a correct ensemble.  Use
> Parrinello-Rahman.
>
> -Justin
>
> --
> ==================================================
>
> Justin A. Lemkul, Ph.D.
> Ruth L. Kirschstein NRSA Postdoctoral Fellow
>
> Department of Pharmaceutical Sciences
> School of Pharmacy
> Health Sciences Facility II, Room 629
> University of Maryland, Baltimore
> 20 Penn St.
> Baltimore, MD 21201
>
> jalemkul at outerbanks.umaryland.edu | (410) 706-7441
> http://mackerell.umaryland.edu/~jalemkul
>
> ==================================================
>
> --
> 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.
>


More information about the gromacs.org_gmx-users mailing list