convergence and stability of simulation

From: Jorgen Simonsen (jorgen589_at_gmail.com)
Date: Mon Aug 15 2011 - 14:08:29 CDT

Hi all

I am trying to simulate a membrane system with a protein embedded but
I a, not quite sure how well my system is behaving.

I would like to monitor some kind of convergence parameter - I have
chosen rmsd of all atoms, total energy, and the volume of the box.

I would expect all parameters to fluctuate around some value after X
ns simulation - for the rmsd the simulation is still increasing even
though it is a very low increment - is this normal or should a expect
a "perfect" normal distribution of the rmsd values? I have tried to
hypothesis test different blocks from the simulation and depending on
the size I get different confidence intervals - is this a known issue?
The same for the total energy - how much variance is normally observed
in simulation of say 50.000 atoms - my variance is a little bigger
than 1/sqrt(N) - I am not experienced with membrane system and I am
not quite sure and much variance to expect.

Any comments/experience??

Thanks

This archive was generated by hypermail 2.1.6 : Mon Dec 31 2012 - 23:20:42 CST