[gmx-users] radial distribution function

Justin A. Lemkul jalemkul at vt.edu
Sat Sep 10 02:34:05 CEST 2011

Moeed wrote:
> On Fri, Sep 9, 2011 at 4:18 PM, Justin A. Lemkul <jalemkul at vt.edu 
> <mailto:jalemkul at vt.edu>> wrote:
>     Moeed wrote:
>         Thank you Justin for your reply. Simulation is 5 ns long (
>         before this a 500 ps NVT was done as equilibration) and total
>         energy has been equilibrated after around 3 ns.
>          -b is set to 4000 that is the last 1 ns has been used for rdf.
>     For all but the simplest systems this simulation length and data
>     analysis period is insufficient.  How big is your polymer?  Proper
>     sampling may take tens to hundreds of ns of data collection.
> Thanks. There are about 250 backbone carbons on 4 chains. Molecular 
> weight of each chains is around 3700 g/mol. I have been looking at total 
> energies and densities to equilibrate by checking sign change of 
> Tot-Drift for these properties before 5 ns. I actually dont see a big 
> change in energies. Is there any criteria to check on this using RMSD or 
> error? (so far I have been using sing change of Tot-Drift).
> Please guide me. Thank you.

The criteria should be whatever experimental observables you wish to reproduce. 
  Once those have converged to reliable values, you've done enough simulation. 
Total energy should always remain (relatively) constant throughout a simulation; 
it is a relatively insensitive measure for assessing equilibrium.

Even simple, robust protein systems are not reliably converged in 5 ns.  I 
sincerely doubt that large polymer chains are either, especially with the 
possibility of entanglement or configurational changes.



Justin A. Lemkul
Ph.D. Candidate
ICTAS Doctoral Scholar
Department of Biochemistry
Virginia Tech
Blacksburg, VA
jalemkul[at]vt.edu | (540) 231-9080


More information about the gromacs.org_gmx-users mailing list