[gmx-users] number of windows in PMF

Justin A. Lemkul jalemkul at vt.edu
Tue Jul 12 23:59:44 CEST 2011

Justin A. Lemkul wrote:
> Rebeca García Fandiño wrote:
>> Thanks a lot for your answer.
>> My system have about 29000 atoms, and the simulation time was 1ns.
>> I have generated the error bars, and indeed you were right, they are 
>> too big. The histograms looked good, so I thought they should be 
>> well-converged...:S
>> What should I do, extending the simulation or generating more windows?
> The simulations probably need to be longer.  I've gotten good 
> convergence in much larger systems using far fewer windows but somewhat 
> longer simulations.

Another thing to consider (aside from my offhand guesses of simulation length) 
is that if you're generating velocities at the beginning of each simulation in 
each window in the absence of prior equilibration, you're not *really* getting 1 
ns of viable data.  Some should be discarded as equilibration, if you're not 
doing that already.  If you've done prior equilibration in each window, then 
disregard this thought.



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